Month: November 2021

Transcript

Introductions [00:20]

Good day folks. This is Shane Hastie for the InfoQ engineering culture podcast. Today I’m sitting down with Carol McEwan and Greg Woods, Carol and Greg are from iObeya. First of all, welcome Carol and Greg. Thanks for taking the time to talk to us.

Carol McEwan: Thanks for having us.

Greg Woods: Yeah, thanks for having us.

Shane Hastie: What’s this Obeya thing? It’s a term we’re starting to see bandied around a bit. But for the benefit of our audience who might not have come across it or who are just seeing the word. What does it mean? Where is it from?

What is an obeya space? [00:53]

Greg Woods: Obeya comes from the Japanese term for large room. And this large room concept was developed by Toyota in the nineties, as they were developing the Prius, they were looking for a new way of visualizing work, visualizing communication, and understanding that they had to come up with something very innovative in a shorter time span and under a budget crunch as well. They were looking to find a new way of working and so they develop the Obeya space and sometimes folks refer to that as the war room and it’s a way of visualizing work so that people can stand in front of the wall or stand in front of the board and have a conversation about what’s happening there. So it’s a way of improving and accelerating dialogue about what needs to get done.

Shane Hastie: How does that differ from something that many of our audience will be familiar with the scrum board or the Kanban board?

Carol McEwan: Yeah, so it’s not a whole lot different, it’s just walls across an enterprise. It takes it up a notch, right. So we know at a team level a scrum board, or we know the team level, a Kanban board. But think about, as you scale, how do you know how you have work dependencies across a team or a team of teams. How can you then start to visualize that work to know, am I taking on too much or am I taking on too little or, oh, I didn’t know that team over there was working on X when I’m actually working on it as well. So maybe we can work together on that. Think about scaling your practices beyond just the board for your team.

Greg Woods: The Obeya may include other walls, such as a visualization of, if you’re looking at the enterprise level, all a visualization of a strategic plan and some of the KPIs that are associated with that and then, as Carol was talking about, getting more detail into the execution of work, Kanban also comes from the Japanese terminology of card. It’s a trigger. It’s a visual cue to know essentially when to move work from one place to another, or when in the case of physical space, a bin might be empty and needs to be refilled or a bin is full and needs to be moved to the next workstation. So whether you’re working in the physical space with a card system like that or virtually with a Kanban board or a scrum board, it’s that visual cue to make progress, to continue executing the work that needs to get done.

Shane Hastie: Greg, you mentioned in the conversation we had before, that your background in particular is lean. So how does this Obeya support lean thinking and lean ways of working in organizations?

The application of lean ideas through obeya [03:43]

Greg Woods: That’s a great question. So lean at the heart of it is ensuring the customer gets what they want when they want it and at the price that they want it. So it’s all about connecting the dots and creating a seamless workflow from the point of your supplier, to your customers. So your classic kind of SIPOC diagram, suppliers, inputs, outputs, processes, and customers. And so this is all connected through visual cues. As I was mentioning about the Kanban, these are the triggers of how do we make sure work progresses as efficiently and effectively as possible.

Shane Hastie: And Carol we’ve known each other for a while, but we’ve come from a very strong agile background. Is there anything specific from an agile perspective that would add to that explanation?

Obeya supports agile approaches [04:31]

Carol McEwan: Absolutely. I’m learning a whole lot more from Greg since I joined iObeya, and then from my knowledge of the agile space, I think what we’re finding is this difference between business agility or doing agile and being agile. Also what practices you are using as a practitioner in the work that you’re doing, those lines between agile and lean are getting a little fuzzier as are going along, I believe Kanban is a great example because some people don’t know in the agile space that Kanban came from lean, right? And so there’s a lot of practices that we’re doing in the agile space that came from lean that have come across to agile. And I see that same thing happening with the Obeya rooms and being able to tie that understanding for the individual to see how is the work that I am doing adding value to the organization’s expectations or their strategy.

If I make this decision down here, I know based on what I’m seeing in the Obeya room, that that decision will trickle all the way up and add value, or not. Sometimes decisions change and in agility we want to be agile, right? We want to be able to make decisions fast and change appropriately as the need arises. And this is an opportunity in an Obeya room to do that. And that’s what excited me so much about joining this organization was because of not looking at lean or not looking at agile so much, but looking at how can we be effective as an entire enterprise and have transparency from top to bottom and bottom to top,

Shane Hastie: I can see a room full of various boards helping but what does it take at a cultural perspective to make that happen?

The culture changes needed to make this effective [06:22]

Greg Woods: That’s a great question. I think Carol hit on it just a moment ago and it’s the notion of transparency as once you stick something up on a wall, it’s there for people to see whether it’s in the physical space or in the digital space. So from a cultural perspective, certainly leaders need to be comfortable with the fact that they’re sharing information that maybe they weren’t sharing before. And by sharing some of that information, they may get challenged on the information, or people may ask questions about decisions being made. So there’s a degree of comfort or confidence or both that leadership needs to have when they start to share information more openly like that. And it brings a greater degree of accountability, not only for the person say whether they’re part of the project team, trying to advance a task on a project or a developer writing a piece of code, but it’s a two-way street.

So leaders are going to be held accountable for, of things on their boards as well. And they’re visible and they’re out there. So from a cultural perspective, really being able to have that comfort and confidence to be able to stand in the space of, we’re putting it all out there and it’s transparent and we need to be able to have open and honest conversations about those things. So Key skills are your active listening skills, being able to give effective feedback, listening to effective feedback, and all of those pieces play into the kind of culture that transforms an organization when you make things transparent throughout the organization.

Carol McEwan: Yeah, and I couldn’t agree more. And I also think the other big thing is trust, right?

Greg Woods: Absolutely.

Leadership buy-in and support is not enough – active leadership engagement is crucial [07:59]

Carol McEwan: Trust huge. And I like to go back to, the state of agile report, right? And if you take a look at the state of agile report, the culture is always a big thing. Culture is king when you’re trying to do this kind of stuff, but they said that in the 14th state of agile report that a new choice this year is that there’s not enough leadership participation in order to help with this. So I think you don’t need leadership buy-in and support. You need leadership participation and engagement. And I was so excited when I saw that report because I know it’s true. We can’t say do as I say, it’s do as I do. So if you’re going to have a good, transparent wall of walls to walk, everybody has to participate in that. And it does open up that transparency and it just brings to light the decisions that are being made throughout the organization. So I think mind shift, change, mindset change. That’s what’s going to help us get there too.

Greg Woods: And I think we need to remember that culture is an outcome. It’s a collective outcome of the behaviors of the people in the organization. I always wonder about initiatives that are out there to “change the culture”. When the initiative is really about changing the behavior. And then however you change the behavior, whatever behaviors are enforced or reinforced or encouraged, that’s the culture that you’re going to get. Culture is an outcome. It’s not just a magic kind of state of being. It’s what you do day in, day out. It’s how you talk to your people day in, day out. It’s how leaders treat their people day in, day out. That’s what culture is. And that’s where it comes from.

Shane Hastie: Johan Rothman has said to me on more than one occasion, that culture is the lowest common denominator of the behaviors that you accept in the organization. What are the behaviors that we need to bring in to start to embody this transparency, to make something like an Obeya effective as a communication tool?

Behaviours that reflect and influence culture [10:02]

Greg Woods: I think one of the classics that you go back to is Stephen Covey-ism of “seek first to understand”, instead of judging what you see on the wall seek to understand why is that on the wall, help me understand that what that card is telling me and seeing if a card is still in the to do column instead of in the progressing column, or if it’s got data in there that says it’s “late”,  jumping to the conclusion of it’s late and seeking a punishment behavior mindset, shifting to the, ask the question first, seek first to understand, why is that card late? Or why is that card not progressed? And more specifically from a leadership perspective, it’s asking the question, if something’s not going, like it’s expected to, what can I do to help? What support can I provide in order to help my team progress. Moving from that place of sitting in judgment of why something hasn’t happened to the place of support to help make sure that it does happen.

Carol McEwan: Couldn’t agree more, Greg, I like to tell a quick story about a BVIR (a big visual information radiator). That’s really also how in the agile space, we might suggest similarity to a visual management system or an Obeya room. But I was part of an opportunity where we were taking agility into a school and teaching children and teachers. And this was a school that these kids needed a little bit more support than maybe a mainstream channel school. And they thought they put the marks up on the wall. Well, let me tell you that didn’t go down very well. However, when they stopped and thought about it, and they iterated and they inspected and adopted to the situation. What they did for the next semester was they put a BVIR, or they put all marks up on the wall, but they started every single child at exceeds expectations.

They all started at a hundred percent. So now what happened when a child, hey I’m a hundred percent and I’m doing great. And I don’t want that slip. And I don’t want anybody else to see that slip. So what happened was some children did obviously, but rather than the other kids berating them for having bad grade, they said, gosh, what happened? How can we help? Hey, I’m good in math. Can I help you? Hey, I’m good in something else, can I help you? And it became a team effort and they swarmed to help each other. And I see that same thing happening in teams in this situation too. Iinformation is not there to berate people. It’s so there to figure out where we can help the best in situations. So make sure we’re providing the right information. It’s not all the information, it’s the right information, so that we can make the right decisions to support the right people at the right time to get the right value. At the end of the day.

Greg Woods: I love that story that makes me think of one of the lean truism out there or phrases that we often hear is, you attack the problem or you attack the process, not the person you go after the piece that is objective. I have objective data about how my process is performing or marks in this case, and you’re not attacking the person, but you go after the process that’s been created, that’s an outcome that isn’t the outcome that you want. So let’s go look at that. Let’s not look at Joe, the operator or Sally, the student let’s figure out how do we really go after that, that we can attack, what can we change.

Carol McEwan: Right.

Shane Hastie: Building on that. There’s something intriguing there in the way that information was communicated. So there was an interpretation layer on top of the raw data. How, as leaders, do we figure out the right interpretation for that?

The importance of representing the data in a way that is useful and meaningful [13:44]

Carol McEwan: I think Greg hit on it a little bit ago. We can all jump to conclusions, but we have to seek to understand. We have to learn. We all have our own personal biases about things. And when we see one thing, we might jump to conclusion of something else. But when the information is presented, it’s only an opportunity to have conversation. And you cannot say based on that data, this is it. You have to be able to use these Obeya rooms as an open opportunity for conversation to be had. And that’s the power of Obeya rooms and opening up that level of communication. Because I think if you go back to the state of agile report, but I think one of the other things that’s an issue is communication between leadership and the people doing the work, or when you are trying to align strategy with execution, how are you doing that? How are you communicating those things and making sure that it’s only a level to open up a line of communication.

Greg Woods: And I think from a visual management perspective, that communication is when you take complex data or complex information and you create a simple visualization out of it. I like to say that the graph is the start of the conversation. When you look at the trend line and if it’s going up and up is good, then starting the conversation with what do doing well that’s got our trend line going in the right direction, or conversely, if the trend line is going down and that’s a representation of, that’s not exactly what we look to be achieving, what’s happening here? What’s happened over time. Let’s go investigate, what’s driving performance in the wrong direction. So being able to build on what Carol was saying, communication, and can start with that visual representation and that visual representation is the trigger for the conversation that needs to be had. How do we keep getting better? Or how do we course correct and get back to where we need to be?

Shane Hastie: So we’ve been talking about the physical Obeya spaces. iObeya is a virtual representation of that. How does this translate into the virtual space and how do we get the same feeling of absolute transparency that you get from being able to walk into a room and look around the walls? How does this translate into the virtual world that so many of us work in today?

Working with virtual Obeya spaces [16:07]

Greg Woods: That’s a great question. And that really strikes at the heart of how the company got started. A major auto manufacturer came to us and said take our flip charts and our sticky notes and put it into a digital environment. And the premise of how iObeya is structured is taking and basing the digital interaction on rooms. We call them rooms and boards. So you can step into a room or you can have permission to access a particular room. And then the boards are much like you’d see either on a wall or the flip charts that are hung in the physical space. And we have dividers that you can envision a corner of the room is a different color. So your left hand wall is on the left hand side of your room, and you’ve got your strategy.

And then there’s divider, which could mimic your physical corner of the room, and then you go into the next section of the room. So we’ve tried to create a structured space, just like the physical space. So it’s much, much easier to adapt and adapt into that digital space because the essential look and feel is like the physical space.

Carol McEwan: And one thing that I’d love to explain too, is the fact that you could think about the advantages of having a digital room over a physical room, right? Like sticky notes don’t fall off the wall, or you don’t have to roll up your program board because you got to move it to another room because you got too big. You could expand that as well. Or now you don’t even have to rewrite a card because you could synchronize a card and share them across rooms. So the value that this virtual room brings, it’s fun to work in too. So it’s unique, it’s flexible. And your imagination is the only thing that limits you sometimes because it is it’s really fun to work in. So it brings an added element to the opportunity.

Asynchronous collaboration over time and distance [18:02]

Greg Woods: I think another couple of key points there to build on what Carol is saying is, is when you’re in that space, it’s the ability to collaborate. Not only at the same time. So Carol and I could be working in the same digital space at the same time or since we’re located in different time zones. And I’ve got a client that I’m working with, that they have a facility on the West Coast (USA), a facility in South Korea and a facility in Singapore, obviously different time zones, but they’re all in the same digital space so they can work on boards. What we like to call is asynchronously.

So if the folks on the west coast make a change to the board, when the folks wake up the next day in Singapore, they can see the changes and they can continue work. So it’s a fantastic way of being able to engage your top talent in the organization, no matter where they are on the face of the earth, they can log in from New York. They can log in from Denver. They can log in from Singapore. They can log in from Paris. And if you can manage all those time zones to get folks together at the same time, you can do that. Or you can literally pass the work around the world and people can be in the same space, working on whatever needs to get done. And quite frankly, the work never has to stop because it’s always there. It’s always present. It’s always live and you can always interact with it.

Shane Hastie: If we can take this to an example that will hopefully resonate with our audience. If we think of an organization that’s got a number of software engineering teams working either in parallel on single products or on complimentary products, streams of work, what would an Obeya space look like for them? What are some of the things that would be up on those walls, whether they’re virtual or physical?

What an Obeya space for a software organisation could look like [19:45]

Greg Woods: The basic foundational piece is the board. So whether, physical, we might equate that to like a flip chart or maybe a whiteboard in our space. That’s a digital board as well as we have your basic sticky notes or what we also like to call cards that have more detailed information in them. So a description of the task, an owner, a due date, maybe a checklist of things that need to get accomplished. And not only do we have the capability of putting up these sticky notes onto a blank board, but we can have both what we call a static background, where you could have different images on there. So just as an example, the first one pops to mind is if you want to do an empathy map with your team, we can create the static background. We have a background catalog that includes an empathy map.

And so boom, right there got your map to work with. And then you can operate on top of it. As we talked about before the same sort of process you can engage with Kanban boards and we have Kanban boards that are dynamic so that we can actually count the number of cards in a particular column, whether it’s your, to do in process or done, however you structure your Kanban and set up a particular amount of whip in each of these columns as well. So if you only want five tasks in your to-do column and you’ve got six tasks listed that whip box or that whip visualization turns red. So you’ve got a number of different ways that you can actually interact with and design your workflows and design how you’re interacting with your information. And you can even create custom backgrounds as well. So it’s not only what’s available in the existing catalog, but if you have a particular way that you like to manage your work or your workflows, those are available to be customized for yourself.

So there’s a whole host of different ways that you can manage your teams, your projects, your work. We like to say that we’re a framework. We don’t say that you have to do lean a particular way and you have to use these boards in this order and do it this way. That’s not our design. Our design is a framework with a powerful toolkit. Then you’re able to manage the work, how you want to. So what would something look like? Imagine how you’re doing it today, just doing it in a digital space.

Carol McEwan: Or your retrospectives or problem solving workshops.  Also the voting, the voting is really cool that I’ve seen, different ways of doing it anonymously and those kind, you get to vote. But I remember doing some problem solving workshops where people were always afraid to vote on something if their leader had gone up and voted already. So the opportunity to vote anonymously and then see all of them collected all at once and things like that even makes it that much more powerful. But like Greg said too, just the way that became was to say, let’s just take everything that was analog and make it digital first and then make it better from there. But those are some of the things that I think make it special.

Using virtual spaces helps encourage diversity and inclusion [22:38]

Greg Woods: I appreciate what Carol’s saying there too, from a voting perspective and what that means to the group dynamic and how do you bring out different ideas? And so I think we’re just touching on the tip of the iceberg here, when it comes to, how do we more effectively facilitate conversation? How do we bring more diversity and inclusion into the digital space? If you’re not seeing what the other person looks like or what their position is in the organization, you can certainly be much, much more inclusive by saying, we’re all contributing here. It doesn’t matter where you’re from, what language you speak, what you look like, however you look to slice it. I think there’s an element of being more powerful or being able to create a more powerful environment that really contributes to diversity and inclusion as we move forward.

Carol McEwan: And the one other thing that I’d like to mention too, is you really get to, to see the incremental improvement that you’ve made along the way, because now you have a record, a source of truth that is digital now that you can keep and you could see the improvements across, whereas before you just kind of know, oh yeah. Remember how it used to be back in the day. Now you can actually look at this and say, look at that. Look how far we’ve come. You know, last year we were here and now we’re here and that’s pretty powerful. I think as a team to celebrate those successes that they’ve made. And we all know that’s what we makes people want to get up and go to work every day to feel like their part of the team and part of the success and the value that they’re bringing to the organization.

Greg Woods: I really appreciate what you’re saying there, because that brings us back full circle to the culture conversation. Celebration and recognition are keys to creating the type of culture, positive environment that organizations need and want to create to help make sure that they’re engaging their folks and getting the best out their folks.

Shane Hastie: Thank you so much. Some really interesting points and some good advice in there. Carol, Greg, if people want to continue the conversation, where do they find you?

Carol McEwan: That’s a great question. They can find cmcewan@iObeya.com.

Greg Woods: And I’m just gwoods@iObeya,com and certainly folks are welcome to check out our website, which is of course just www.iobeya.com.

Shane Hastie: Thanks so much.

Carol McEwan: Thank You.

Greg Woods: Thanks for having us, Shane.

Mentioned

Prominent companies are thriving in the enterprise software space due to continued digital transformation, adoption of hybrid lifestyles, and a growing market for it. Use Intuit (INTU). MongoDB Their solutions should be in high demand (NASDAQ:). Which stock is better to buy right now? Find out more. Intuit Inc., NASDAQ:, provides compliance and financial management products and services worldwide for small and medium-sized businesses and individuals. The Mountain View, Calif., company operates in four segments: Small Business & Self-Employed; Consumer; Credit Karma; and ProConnect. MongoDB, Inc., New York City (MDB), provides a global general-purpose database platform. MongoDB Enterprise Advanced (NYSE:), MongoDB Atlas (NYSE.) and Community Server are some of the products offered by this company. The company also offers professional services such as consulting and training.

Threats related to data security, especially on cloud-based platforms, continue to hamper the enterprise software market’s growth. However, there is still a strong chance that the market for enterprise software will grow quickly in the months ahead due to increased demand from nearly every industry. This comes as part of the wider digital transformation effort. A resurgence within COVID-19 cases is causing a revival in hybrid working arrangements. This is proving to be a benefit for enterprise software. A Statista report states that the global enterprise software market will grow by 8.74% between 2021 and 2026. Therefore, INTU and MDB both should be benefited.

INTU’s shares have gained 15.8% in price over the past month, while MDB has returned 0.5%. Also, INTU’s 65.6% gains over the past nine months are significantly higher than MDB’s 27.4% returns. Furthermore, INTU is the clear winner with 80.1% gains versus MDB’s 41.3% returns in terms of their year-to-date performance.

Continue reading on StockNews

Disclaimer: Fusion MediaWe remind you that this site does not contain accurate or real-time data. CFDs are stocks, indexes or futures. The prices of Forex and CFDs are not supplied by exchanges. They are instead provided by market makers. As such, the prices might not reflect market values and could be incorrect. Fusion Media does not accept any liability for trade losses that you may incur due to the use of these data.

Fusion MediaFusion Media and anyone associated with it will not assume any responsibility for losses or damages arising from the use of this information. This includes data including charts and buy/sell signal signals. You should be aware of all the potential risks and expenses associated with trading in the financial market. It is among the most dangerous investment types.

Transcript

Introductions 

Srini Penchikala: Hi everyone. My name is Srini Penchikala. I am the lead editor for AI/ML and data engineering community at InfoQ website. Thank you for tuning into this podcast. In today’s podcast, I will be speaking with Meenakshi Kaushik, and Neelima Mukiri, both from Cisco team. We will be talking about machine learning algorithm bias, and how to make machine learning models fair and unbiased.

Let me first introduce our guests, Meenakshi Kaushik currently works in product management team at Cisco. She leads Kubernetes and AI/ML product offerings in the organization. Meenakshi has interest in AI/ML space and is excited about how the technology can enhance human wellbeing and productivity. Thanks, Meenakshi, for joining me today. And Neelima Mukiri is a principal engineer at Cisco. She is currently working in Cisco’s on-premise and software service container platforms. Thank you both for joining me today in this podcast. Before we get started, do you have any additional comments about your research and projects you’ve been working on, that you would like to share with our readers?

Meenakshi Kaushik: Hi everyone. My name is Meenakshi. Thank you, Srini, for inviting Neelima and me to the podcast. And thank you for that great introduction. Other than what you have mentioned, I just want to say that it is exciting to see how machine learning is getting more and more mainstream. And we see that in our customers. So, this topic and this conversation is super important.

Neelima Mukiri: Thank you, Srini, for that introduction and thank you, Meenakshi. Definitely we are very excited about the evolution of AI/ML, and especially in the Kubernetes community. How Kubernetes is making it easier to handle MLOps. I’m excited to be here to talk about fairness and reducing bias in machine learning pipelines, as ML becomes more pervasive in the society. That’s a very critical topic for us to focus on.

Srini Penchikala: Thank you. Definitely I’m excited to discuss this with you today. So, let’s get started. I think the first question, Meenakshi, maybe you can start us off with this. So, how did you get interested in machine learning, and what would you like to accomplish by working on machine learning projects and initiatives?

Meenakshi Kaushik: Machine learning has been there for a while. What got me excited is when I started seeing real world use cases getting more and more deployed. So, for example, I remember a much change when I saw Amazon’s recognition, and it could recognize facial expression, and tell you your mood. What I took back from that is, “Oh, isn’t that helpful? You can change somebody’s mood by making them aware that today you’re not looking so happy.” So, that was pretty groundbreaking. And then more and more applications came along, especially in image recognition, where you could tell about patients’ health, and that became more and more real. And we, as Neelima pointed out, that got mainstream even with our customers, with the evolution of Kubernetes and Kubeflow. So, both these spaces together, where it became easier and easier to enable data scientists and even ordinary folks to apply machine learning in day to day, really got me excited. And this evolution is progressing, so I feel very happy about that.

Srini Penchikala: How about you, Neelima, what made you get interested in working on machine learning projects, and what are your goals in this space?

Neelima Mukiri: I’ve always been interested in AI/ML and the possibilities that it opens up. In the recent past years, advances in ML have been so marked compared to 10 years before. There’s so much improvement in what you can do, and it’s so much more accessible to every domain that we are involved in. The possibilities of self-driving cars, robotics, healthcare, all of these are real world implications that have a chance to affect our day to day lives. In addition to just how exciting the field is, being involved in Kubernetes and in Kubeflow as part of Cisco’s container platforms, we’ve seen our customers be very interested in using Kubeflow to make ML more accessible. And as part of that, we’ve started working on AI/ML in the context of Kubernetes.

Define AI fairness and examples of fair AI solutions

Srini Penchikala: Yeah. Definitely, kubernetes brings that additional dimension to the machine learning projects, to make them more cloud native and elastic and performant, right? So, thank you. Before we jump into the machine learning bias and fairness, which is kind of main focus of your discussion here, can you define AI fairness? What does it mean by AI fairness? And talk about a couple of examples of fair AI solutions, and an example of ML solution where it hasn’t been fair.

Meenakshi Kaushik: The fairness comes into the picture when you start subdividing the population. So for example, this is an example which we gave even in our KubeCon presentation, where let’s say a bank is looking at giving loans to the entire population. And it decides that 80% of the time it is accurate. So, overall in the population, things behave normally. But when you start looking at subsection of population, you want to see whether the subsection of population are equally getting represented in the overall decision making.

So, let’s say if 80% of the time, the loan application gets accepted, if you started slicing and dicing at a broad level between, let’s say, male and female, 80% of the time equally, do they get accepted? Or within the population where people who had previous loans, but defaulted, whether they get equally represented or not? So, fairness is about looking at a broad solution and then slicing and dicing into a specific subgroup, whether it is based on gender or racial differences or age differences. For example, with COVID vaccine, if it was tested on adults, it doesn’t work on children that well. So, it’s not fair to just push your machine learning data to children until you have looked at that population, and it’s fair to that population. So, fairness is about equity, and it’s really in the context of the stakeholder. A stakeholder decides at what level they want to define fairness, and what groups it wants to figure out, whether it is fair across or not.

Three sources of unfair algorithm bias

Srini Penchikala: That’s a good background on what is fair and what is not, right? So, maybe Neelima, you can probably respond to this. In your presentation at KubeCon conference last month, you both mentioned that the sources of unfair algorithm bias are data, user interactions, and AI/ML pipeline. Can you discuss more about these three sources and the importance of each of them in controlling the unfair bias and contributing to unfair bias?

Neelima Mukiri: So, data is the first step where you are bringing in real world information into a machine learning pipeline. So, let’s say you take the same example of deciding whether a person is creditworthy for a bank to give a loan or not. You populate the machine learning pipeline with data from previous customers, and the decisions made for the previous customers, whether to give them a loan or not. So, the data comes in from the real world, and it is filled with bias that is present in our real world because the world is evolving and what we consider as fair a few years back, may not be fair today. And we are all prone to prejudices and biases that we put into our decision making process. So, the first step in the machine learning pipeline is that collection of the data and processing the data, where we have the chance to evaluate and see what are the biases that are present.

For example, are there large subsets of the population that are missing in the dataset that we have collected? Or is the data skewed towards being more positive for some subset of a population? So, that’s the first step where we want to evaluate, understand and when possible, improve the bias that is present.

The next step is, in the machine learning pipeline as we build the model and as we are serving the model, every step of the pipeline we want to make sure that we are considering and evaluating the decisions that are made in the pipeline, the models that are built and the inference that is provided, to evaluate and see, is it being fair across the population set that you’re covering? And when you bring that decision to a user and present it, that can in turn reinforce bias by the user acting on it.

So, let’s say, you’re looking at policing and you are giving a wrong prediction of somebody being prone to commit a crime. It’s possible that the police will actually do more enforcement in that region, and that ends up assigning more people in that region as possible to create a crime, and then that feeds into your data and the cycle is reinforced. So, every step in the pipeline right from starting from where you collect your data, to building models, providing inference and then seeing how people act based on those inferences, is prone to bias. So, we need to evaluate and correct for fairness where possible.

AI/ML fairness toolkits

Srini Penchikala: Maybe, Meenakshi, you can respond to this one, right? So, can you talk about some of the AI/ML fairness toolkits that you talked about in the presentation, and why you both chose Kubeflow for your project?

Meenakshi Kaushik: As we were talking in the beginning of the presentation, we work in the Kubernetes domain. And although there are many machine learning lifecycle manager toolkits available on top of Kubernetes, Kubeflow has gained a lot of traction, and it’s used by many of our customers. It’s also pretty easy to use. So, we chose Kubeflow since it is one of the popular open source machine learning lifecycle manager toolkit. And really, what it allows you to do, it allows you to build all the way starting from your exploration phase into the production pipeline. It allows you to do everything. You can bring up a notebook and run your machine learning models, and then chain them together in a workflow and deploy it in production. So, that’s why we used Kubeflow.

And then on the machine learning toolkits, when we started this journey, we started looking at open source toolkits. And fairness is an upcoming topic, so there is a lot of interest, and there are a lot of toolkits available. We picked the four popular ones because they had a wide portfolio of features for fairness available. And the good thing is that they had many commonalities, but they also had interesting variations, so that it gives you a large variety of toolkits. So, let me quickly talk about the four toolkits. We started by looking at Aequitas. Aequitas fairness toolkit, I would say is the simplest toolkit when you want to get into. You just give your predictions and it will tell you about fairness. It would give you your entire fairness report. So, your prediction and your data and which population you want to look at fairness, the protected group, and it’ll just give you the data. So, it offers you an insight as a black box, which is pretty nice.

But what if you want to go next level deeper, or what if you wanted to do interactive analysis? In which case, what I found was that Google’s What-If Tool, was pretty nice. In the sense that it is a graphical user interface, and you can do interactive changes to your data to see when it is fair and whether, “Can I get a counterfactual? Can I change the threshold to see if it changes the bias in this subpopulation?” And how it impacts other things. For example, it might impact your accuracy if you try to change your bias threshold. So, What-If Tool is pretty good from that perspective. It is interactive and it will help you with that. Obviously, because it’s an interactive toolkit, if you have billions and billions of dataset, you won’t be able to pull all of those into this graphical user interface. But there is some strength to having a graphical toolkit.

Then the other toolkits which we looked at are AI 360 degree from IBM, and Microsoft’s Fairlearn. And these toolkits are awesome. They don’t have the interactive capability or a white box capability of Aequitas, but they have very simple libraries that you can pick and put in any of your machine learning workflow, on I guess, any notebook. In the case of Kubeflow, it’s Jupyter notebook, but you could definitely run it on Colab. And now as you are experimenting, you can see graphically using those libraries where their fairness criteria lies.

So, those are four toolkits, and all of these toolkits have strength in doing binary classification, because that’s where the machine learning fairness toolkits have started. For other areas like natural language processing and computer vision, things are evolving. So, these toolkits are adding more and more functionality into it. So, that’s an overview of the landscape that we looked at.

Srini Penchikala: Neelima, do you have any additional comments on that same topic? Any other criteria you considered for the different frameworks?

Neelima Mukiri: In terms of the toolkits, Meenakshi covered what we looked at primarily. And in terms of the criteria, one of the primary things that we were looking for was how easy is it to run these on your on-prem system versus having to put your data in a cloud. Given a lot of our customers are running their workloads On-prem and they have the data locality restrictions. That was one key thing that was important for us to understand. And all the toolkits we were able to run them on-prem in Kubeflow. Some of them are, especially What-If, is a lot easier to run directly. Go on to the website and run it in a browser, but you have to upload your data there. The other part that we looked at is the programmability or how easy is it to bundle this into a pipeline? And that’s where, I think, both Fairlearn and IBM AI 360 are easier to plug into, as well as a bunch of the TensorFlow libraries that are available for bias reduction and detection as well.

Yeah. So, the two axes which we were coming from was, how easy is it to plug in your data to it. And then where can you run it. How easy is it to run it in your machine learning pipeline versus having to have a separate process for it?

Srini Penchikala: So, once you chose the technology, Kubeflow, and also you have the model defined, and you have the requirements finalized, so the most important thing next is the data pipeline development itself, right? So, can you discuss the details of the data pipelines you are proposing as part of the solution to detect the bias, and improve the fairness of these programs, right? So, there are a few different steps you talked about in the presentation such as pre-processing, in-processing and post processing. So, can you provide more details on these steps? And also more importantly, how do you ensure fairness in every step in the data pipeline?

Neelima Mukiri: Basically, we divided the machine learning pipeline into three phases. In-processing, pre-processing, and post processing. Pre-processing is everything that comes before you start building your model. In-processing is what happens while you’re building your model, and post processing is, you’ve built your model and you’re ready to serve, is there something that you can do at that point? So, the first part, which is pre-processing is where you look at your data, analyze your data, and try to remove any biases that are present in the data. The type of biases that are better served by handling at the stage, are cases where you have a large skew in the data available for different subgroups. The example that we gave in the presentation was, let’s say, you’re trying to build a dog classifier, and you train it on one breed of dogs. It’s not going to perform very well when you try to give it a different dog breed, right?

So, that’s the place where you’re coming in with a large skew in the data available per subgroups, trying to remove it at the pre-processing phase itself. The type of biases that are more easy to remove, or better served by removing in the model building phase, are more of the quality of service improvements. So, let’s say you’re trying to train a medical algorithm to see what type of medicine or treatment regimen works best for a subset of population. You don’t really want to give everyone equal medicine or equal type of medication, you want to give them what is best serving their use case, what works well for that subset. So, you actually want to better fit the data.

And that’s where doing the bias reduction during the model training phase, which is the in-processing step, works better. And there are a bunch of techniques which are available to improve or to reduce bias in the model training stage, that we talk about in the presentation, like going through and generating an adversarial training step, where you’re trying to optimize for both accuracy as well as reducing the bias parameter that you specify.

Now, when we have trained the model, and we’ve decided on the model that we are going to use, we can still evaluate for bias and improve fairness in the post processing step. And the type of data that is really well suited for that is where you have existing bias in your data. So, the example of where you have the loan processing, where let’s say a subgroup of population is traditionally being denied loans even though they are very good at paying back the loan. So, there you can actually go and say, “Hey, maybe their income is less than this threshold, but this population has traditionally been better at paying back the loan than we’ve predicted, so let’s increase the threshold for them.” And you’re not changing the model, you’re not changing the data, you’re just increasing the threshold because you know that your prediction has been traditionally wrong.

So, that’s the post-processing step, where you can remove that kind of bias better. So, each step of the pipeline, I think it’s important to first evaluate and see, and try to remove the bias. And also try different mechanisms, because each thing works better in different scenarios.

Srini Penchikala: Meenakshi, do you have any additional comments on the data pipeline that you both worked on?

Meenakshi Kaushik: Yeah. What happens even before we have the ability to do pre-processing, in-processing or post-processing is, what do we have at hand? So, for example, sometimes we didn’t build the model, we just are consumers of the model. In which case, there isn’t much you can do other than post-processing. Or can we massage the output of the model to become fair? So, in that case, post-processing is and actually, it works very well in many scenarios.

So, it’s not that nothing is lost there, you can still change and make it more fair just by that. Now, sometimes you have access to data, you may not have access to the model. So, in addition to what Neelima is saying about going through the different phases of the pipeline, do not be afraid even if you have a limited view of your infrastructure, or how you are serving your customers. There is still an opportunity where you can massage the data, like at the pre-processing layer.

If you don’t have access to the model, but you have the ability to feed to the model the data, that’s good. You still have the ability where you can change at the pre-processing level to influence the decision. But it’s important to look at what really works. Sometimes the way I look at it is that, ideally it’s like security, you shift left, you try to change the earliest pipeline as possible. But sometimes influencing the earlier pipeline may not give you the best result. But ideally that’s what you want to do.

First, you want to fix the word so that you get perfect data. But if you cannot get the perfect data, can you massage it so that it is perfect? If that’s not possible, then you go lower in the pipeline and say, “Oh, okay, can I change my model?” At times, model changing may not be possible. Then even at the last stage, as we’ve seen in a variety of examples, it’s very good enough where the model may not be fair, but you massage the actual result which you give out to the others by changing some simple thresholds, and make your pipeline fair.

Data pipelines to detect machine learning bias and improve the fairness of ML programs 

Srini Penchikala: Very interesting. So, still I have a question on the fairness quality assurance, right? Neelima, going back to your example of loan threshold, probably increasing it, because it’s been traditionally wrong with the previous criteria. So, how do you decide that, and how do you ensure that that decision is also fair?

Neelima Mukiri: In examples like the bank loan, typically the way to evaluate fairness is, you have one set of data, let’s say from your bank, and the decisions that you have made. But let’s say, you’ve denied a loan to a person, and that person’s gone and taken a loan with another bank. You actually have real world data about how they performed with that loan. Did they pay back on time or not? So, you can actually come back and say, “Hey, that was a false negative that I said, that I don’t want to give a loan to the person, who actually is paying back on time.”

So, historic data, you can take it and see how it’s performed versus your prediction. And you can actually evaluate what is the real fairness in terms of both the accuracy. And you can easily look at fairness in terms of subpopulations by looking at the positive rates per population. But as a business coming in, you want to optimize for value. So, it’s critical to know that you’ve actually made mistakes, both in terms of accuracy, and there is bias there.

The bias is what has induced the errors in accuracy. First of all, getting that historic data, and then getting a summary of how it’s performed across these different dimensions, is the way for you to see what bias exists today. And if you improve it, is it actually going to improve your accuracy as well, and your goal of maximizing profit or whatever your goal is, right? So, tools like Aequitas and What-If, actually give you a very nice summary of the different dimensions. How is accuracy changing as you’re changing fairness? How is it changing when you’re trying to fit better to the data or when you’re trying to change thresholds?

So, I would say evaluate this, run through the system, see the data that it generates, and then decide what sort of fairness reduction that makes sense for you. Because really, it doesn’t make sense to say, “Give it to everyone.” Because you have a business to run end of the day, right? So, evaluate, see the data and then act on the data.

Standards and community guidelines to implement responsible AI

Srini Penchikala: In that example, financial organizations, they definitely want to predict from an accuracy standpoint to minimize a risk, but also they need to be more responsible and unbiased when it comes to the customers’ rights. Okay, thank you. We can now switch gears a little bit. So, let’s talk about the current state of standards. So, can you both talk about, what is the current situation in terms of standards or community guidelines? Because responsible AI is still an emerging topic. So, what are some standards in this space that the developer community can adapt in their own organizations to be consistent with fairness? So, we don’t want the fairness to be kind of different for different organizations. How can we provide a consistent standard or guideline to all the developers in their organizations?

Meenakshi Kaushik: So, let me just start by saying that, as you mentioned, fairness is still in its infancy, so everybody’s trying to figure out. And the good thing is that it’s easier to evaluate fairness, because you can look at lines from the subpopulation and see whether it is still doing the same thing as it’s doing for the overall population as a whole. Given that the easiest thing which you can do for now, which is commonly done in most of our software and even in hardware, is we have a specification. It tells you, “Oh, these are the performance, it will only accept this many packets per second. These are the number of connections it would take.” Things like that. What is a bounding limit under which you would get an expected performance?

And the model now has something called model cards, where you can give a similar specification as to how was the model built? What are the assumptions it made? This was the data it took. This is what the assumption it is making for the bounding limit under which it works, right? This is the data set that it took. For example, if you were doing some kind of medical analysis, and it took a population which is, let’s say from India, then it has a specific view of just a specific population. And if you’re trying to generalize, or if somebody’s trying to use in a generalized setting, a model card which tells you about that, then me as a consumer can be aware of that, and can say, “Aha, okay, I should expect some kind of discrepancy.” Currently, those things are not readily available when you go to take some model from open source or from anywhere, for that matter. So, that’s the first easy step that I think that can be done in the near term.

In the long term, there has to be perhaps more guidelines. Today there are different ways of mitigating fairness. There is no one step which fits all. However, adding those into the pipelines, what needs to be added is not standardized. What should be standardized is that these are the sets of things your model should run against, right?

So, if your model is doing some kind of an age group across all the age groups, then some of the protected groups should be predefined. “Oh, I should look at children versus pre-teens versus adults, and see if it is performing in the same way.” If there is some other kind of disparity, there has to be a common standard that an organization should define, depending on the space they are in. For the bank, for example, it would be based on gender differences, based on the zones that they live in, area zip code they live in, some ethnicity for example. In the case of ofcourse medical, the history is larger. So, those are the near term standards. The broader term standards, I think will take a longer time. Even within machine learning, there are no standard ways to give predictions. You can bring your own algorithms, and you can bring your own things. So, I think we’re a little far away on that.

Neelima Mukiri: Yeah. I would echo what Meenakshi said. We were surprised with the lack of any standards. The field is at it’s very infancy, right? So, it’s evolving very rapidly. There’s a lot of research going on. We are still at the phase where we are trying to define what is required versus at a state where we are able to set standards. That said, there are a lot of existing legal and society requirements that are available, in different settings, what’s the level of disparity that you can have across different populations? But again, that’s very limited to certain domains, certain use cases, maybe in things like, when you are applying for jobs or housing or giving out loans. So, there are fields where there are legal guidelines already in place. In terms of, what is the acceptable bias across different subgroups, that’s where we have some existing standards.

But then bring it on to machine learning and AI, there’s really no standards there. When we looked at all these different frameworks that are available for reducing bias, one interesting thing is that even the definition of what is bias or what is parity, is different across each of these models. Broadly, they fall into either an allocation bias or QoS or a quality of service improvement. But again, each framework comes and says, “This is the bias that I’m trying to reduce, or these are the set of biases that I allow you to optimize.” So again, it makes sense at this stage to actually look at it from multiple angles, and try out and see what works in a specific sub-domain. But as a society, we have a lot of progress to do and ways to go before we can define standards and say, “This is the allowed parity in these domains.”

Future innovations in addressing machine learning biases

Srini Penchikala: Right. Yeah. Definitely, the bias is contextual and situational, and relative, right? So, we have to take the business context into consideration to see what exactly is bias, right? Can you guys talk about what’s the future? You already mentioned a couple of gaps. So, what kind of innovations do you see happening in this space or you would like to see happen in this space?

Meenakshi Kaushik: As Neelima pointed out, we were happy to see that there was fairness. It’s easy to not define fairness, but at least evaluate fairness because it’s model generated rather than, there is some human in the loop involved, and you can’t really evaluate. So, that’s good thing. What I am excited to see is that we are continuing fairness across different domains of machine learning, so that it started with, as I said, classification problems, but it is now going more and more towards the problems which are getting increasingly deployed. Anything to do with image recognition, computer vision, for example, and it touches broad areas, from medical to, as Neelima was pointing out, autonomous driving field. So, that I’m really excited to see.

The second thing is that more and more, hopefully the model cards become the way of the future. Every model comes with what it was that was used to generate the model, and what should the expected output be, so that we all can figure out how it is done. Even for advertisements which are served to me. If I know exactly how the model was defined, it’s useful information to have. So, I’m excited to see that.

And the toolkits which are developing are also very good. Because right now, these toolkits are one-off toolkits. And when Neelima and I started looking at not only Kubeflow, but researching as to what we want to demonstrate in KubeCon, we were looking at a way of automating in our machine learning pipeline. Similar to how we generate automated hyperparameter, we wanted to automatically modify our machine learning model to now have fairness criteria built-in.

So, currently those things are not totally automated, but I think we’re very close. We could just modify some of our routines, similar to the hyperparameter tuning. Now there is a machine learning fairness tuning, so you can tune your model so that you can achieve fairness as well as achieve your business objectives. So, accuracy versus fairness is easily done. So, that’s the other area I’m excited to see that we achieve, so that it becomes in-built like hyperparameter tuning. Also do the fairness tuning for this model.

Neelima Mukiri: Yeah. To echo what Meenakshi said, we really need to have more standards that are defined, that we can use across different types of problems. We also want to see standardization in terms of defining fairness, evaluating fairness. And there’s a lot of improvement to be done in making it easy to integrate fairness into the pipeline itself. There’s work ongoing in Kubeflow, for example, to integrate evaluation of fairness into the inference side of the pipeline, post-processing. And so, we need to be able to build explainable, interpretable models and make it easy for people to build in fairness into the pipelines, so that it’s not an afterthought, it’s not coming in as someone who’s interested in making your model more fair, but it’s part of the pipeline. Just as you do training, testing, cross validation, you also need to do fairness as part of the pipeline, as part of the standard development process.

Final thoughts and wrap-up

Srini Penchikala: Yeah. Definitely, I agree with you, both of you. So, if there is one area that we can introduce fairness as another dimension, and build the solutions out of the box right from the beginning, to be fair, that area would be machine learning, right? So, thanks, Neelima. Thanks, Meenakshi.

Neelima Mukiri: Thank you for this opportunity to talk to you and talk to your readers on this very exciting topic.

Meenakshi Kaushik: Thank you so much for the opportunity. It was fun chatting with you, Srini. Thank you.

Srini Penchikala: Thank you very much for joining this podcast. Again, it’s been great to discuss this emerging topic, and a very important topic in the machine learning space, how to make the programs more fair and unbiased, right? So, as we use more and more machine learning programs in our applications, and as we depend on machines to make decisions in different situations, it’s very important to make sure there is no unfairness as much as possible. To one demographic group or another. So, to our listeners, thank you for listening to this podcast. If you would like to learn more about machine learning and deep learning topics, check out the AI/ML and the data engineering community page on infoq.com website. I encourage you to listen to recent podcasts, and check out the articles and news items my team has posted on the website. Thank you.

Buy low, sell high. It takes money to make money. If you don’t have a lot of cash, these adages could convince you not to invest right now with stock market valuations at a premium.

However, there are still stocks that have tremendous growth potential. And you don’t need a huge amount of money to buy them. If you’ve got $3,000, here are three growth stocks to buy that could skyrocket.

1. Fiverr

COVID-19 opened the door for many individuals and businesses to rethink how they work. This seems likely to accelerate the already growing increase of freelancing. Few companies are as well-positioned to profit from this trend as Fiverr (NYSE:FVRR).

Fiverr’s online platform matches freelancers with buyers of digital services. The company charges a 5.5% service fee to buyers and takes one-fifth of the transaction amount charged by the freelancer. 

This business model has worked really well for Fiverr. The company’s revenue has soared more than 260% over the past five years. Fiverr delivered exceptionally strong third-quarter results with revenue jumping 42% year over year. 

Even better, Fiverr still has a massive growth opportunity ahead. The company estimates that its addressable market stands at close to $115 billion annually. Fiverr currently claims only a fraction of a percent of this market. This stock could easily be a five-bagger or more by the end of the decade.

2. MongoDB

Every time you do anything on the internet, it creates more data. And that data has to be stored somewhere. Increasingly more of the data is stored in the cloud. MongoDB (NASDAQ:MDB) offers what is arguably the best cloud-based database platform around.

Sales for that platform — Atlas — skyrocketed 83% year over year in the second quarter. MongoDB continues to roll out new features that attract more customers. The company’s CEO, Dev Ittycheria, wasn’t exaggerating when he said in September, “It is becoming increasingly clear that the fastest and most compelling way to build modern applications is to use MongoDB.” 

The main knock against MongoDB is that its shares are expensive, trading at nearly 50 times sales. By comparison, none of the so-called FAANG stocks have price-to-sales multiples above 11, with most of them in the single digits. 

MongoDB has much better growth prospects than the FAANG stocks, though. The cloud database market is projected to increase by a compound annual growth rate of 14.8% through 2028. But MongoDB is growing much faster than its rivals with no end in sight to its opportunities.

3. Trupanion

For many Americans, their pets are a part of their family. But veterinary costs continue to rise. That’s where Trupanion (NASDAQ:TRUP) offers great value with its medical insurance for cats and dogs.

Trupanion ranks as the leader in the pet medical insurance market. It had over 1.1 million enrolled pets as of Sept. 30, 2021. The company delivered 40% year-over-year revenue growth in the third quarter. 

Probably the biggest competitive advantage for Trupanion is its relationships with veterinarians. It’s the only company that provides software that allows veterinarians to receive payment for services within minutes after checkout.

Trupanion has solid near-term growth drivers, especially with Aflac offering its pet medical insurance to employers in 2022. Its long-term growth opportunity is even greater. Only 1% and 2% of pets in the U.S. and Canada, respectively, are covered by insurance. Trupanion should be a big winner as it penetrates more of this market.