About This Episode

Episode 10 features Foad Dabiri, a head engineer for Twitter. With a Phd in Computer Science Foad has led teams of people not only in Engineering positions but in CTO, and CEO capacities. Working for healthcare companies like Wanda and Oncoverse, then shifting his focus to social media. Foad brings his unique perspective on team building and leadership. Sharing his top 5 tips on creating an empathetic work culture while learning through experience. 

Foad’s Top 5 Leadership Tips:

Below is a summary of the Top 5 Leadership tips shared during the interview this week. Take a listen to the episode to learn more about the thoughts behind these tips –

  1. Empathy
  2. Is There An ROI On Your Time?
  3. Understanding The Uncertainty And Contradictions
  4. Is Chronic Discontent Slowing You Down?
  5. Measuring Your Impact

We hope you enjoy the episode. You can find even more Full Stack Leader episodes here:


Show Transcript:

Ryan: Hello everyone. And welcome to this week’s episode of the full-stack leader. We’re excited to have Foad Dabiri with us this week. He is a senior manager in machine learning engineering at Twitter, and in the past, he has been a CEO, a CTO at companies like Wanda and Oncoverse. Really excited to have you here. Welcome, Foad

Foad: Thank you for having. 

Ryan: Yeah, we have so many great questions for you. Being a leader in the world of ML there’s a lot of people interested in this area right now and kind of working through some of the challenges, both on the technical side and even the philosophical side, there’s a lot of good spots to talk about. Great. Why don’t you give us a little background on how you got to where you’re at now? 

Foad: Sure thing. So let me start by talking about my graduate studies. I went to UCLA for my PhD in computer science. During the time my focus was. Competent in science theory and algorithm design. And one thing that we did at UCLA was while we were working on theoretical aspects of computer science and algorithms design. The one objective we had was to ensure that we find real applications and apply those learnings into real applications. So I started working on a health care application. That’s UCLA, because there are. Both nursing schools and medical schools down there. And we started working with those colleagues and finding real healthcare applications and how computer science machine learning, and that can help after graduation. I started working at a startup, not in healthcare, but in contextual text processing. So it’s still an exciting domain in how to understand and interpret intent from contextual texts. Shortly afterwards, I joined Google as a software engineer. I was in the search ads, quality, which again, the fundamental problems were quite similar. Then in 2014, The research that he had done at UCLA, which I had remained involved with led to some interesting, and impactful results. At that point, we decided that we would build a company around it. And that’s how Wanda, was born, was a healthcare AI story. With the mission to utilize information we can collect from individuals remotely and use that information to predict adverse events that they might experience in the near future. Let me give you a tangible example. For instance, one of our main application areas was in CHF or congestive heart failure. What we did with it. To collect very simple non-invasive data points from patients at home who had CHF and then use that to predict whether they are likely to experience an adverse event in the next seven days. And we would provide that information to the care team and they would intervene in time. To avoid those adverse events. So that was the premise, of the company. 

Ryan: yeah, real quick on that front, that’s obviously a huge area for machine learning right now. And looking at how you can do those kinds of predictions with simple sets of data. I noticed one of the things you said was, it sounded like it was very basic levels of data. You were able to do the predictions off of, was that kind of a key component that you had to work around as some of the privacy issues in it? 

Foad: Exactly. Especially the simple datasets as well as low volume of data sets. Because a lot of the interesting problems with machine learning involves being able to actually analyze and train using a large amount of data, but in healthcare, especially when you look at clinical trials, the numbers. Participants with the amount of data are much smaller, right? So is this about the traditional big data problem, but more about how to actually best utilize the limited amount of information that you have at the time to come up with a reasonable prediction while the broader hope is that once this becomes more widely used, you will be in a position to actually have a large amount of data.

Ryan: Yeah, you can feed the data back in instead of having servers and servers of data to start, like you really have to, use that very limited set, to make some early predictions and see how they come out. Right. 

Foad: Exactly. 

Ryan: Great. Um, Yeah keep going. 

Foad: Yeah. So, While we were at Wanda, we had the opportunity to actually, create a second company by the name of uncle burrs in oncology, Dwayne, it was a joint venture between us one, and, dignity health, which is one of the prominent healthcare systems in California. And that was focused on enabling. Virtual tumor boards for cancer patients. And again, down the road, trying to utilize the information that is captured in those conversations for more analytical assessment, as you pointed out dry, and the notion of privacy is doubly important if not more in healthcare.So that was its own challenge. There’s HIPAA compliance. Ensuring that dealing with a small amount of data, the privacy of the patients in the privacy of the caregivers are properly maintained. So we worked on a wonderful five years in 2019, , to be, sold the company to an entity in the United Kingdom.

 And one of the reasons. that interested party and it came to our attention because one of our main investors in the UK. So that’s how the networking happened after Wanda. I was quite keen to go back to a more direct to consumer, things that were really important for me were areas that machine learning can play a major role in improving the product experience and serving the correspondent customers. I’ve been very interested in impact and then I believe that. The role a tutored shall talk more about is an area that I am with my teammates and colleagues can be quite impactful. And that’s how I joined Twitter. Twitter’s notifications. 

Ryan: It’s amazing. Yeah. It seems like an absolute kind of gold mine for being able to build tools that can help support some of these impactful changes, and how much it impacts the entire world as we’re seeing.

Foad: Exactly. And then that’s what of the. Bigger appeals of Twitter. Twitter is an application and granted I’m biased. I work at Twitter, but I think it has repeatedly shown that it has a big impact on societies, but, good and bad. There are times that. Tutor as a company is trying to minimize the negative impact that potentially can have, and something that a lot of people rely on to find conversations related to what they care about. So it’s a very exciting application, to try and make a contribution to. 

Ryan: I, yeah, and I think this is an interesting point just in around tech leadership itself. I remember, Going to South by Southwest and the plate two thousand when Twitter was first coming out and it was really the golden child of that particular year of, south by, and 12, 13 years later, as a company. And it’s got a totally different face as an organization and an impact on the world. And you, as a leader within that organization have to carry a lot of that on your shoulders. How does that impact you in the way that you work with. 

Foad: So Twitter has been very intentional about setting clear strategies and initiative. Across the company bought with the new leadership, which is under the Pirog. And before that other Jack and as a leader, the primary challenge is to ensure that we are fully aligned with this strategy and our execution because companies realize the impact and the outcome that our teams provide at the same time, making sure. The individuals, engineers and other cross-functional partners in the team are property excited, motivated, and see that strategy and vision, and actually share those values. Once these two come into place, more often than not the outcomes are what is expected of us. 

Ryan: How much time do you spend working with teams to make sure that they are aligned with those values?

Foad: It’s a continuous process, meaning it’s not like at the beginning of the quarter, we tried to create alignment and then the start to execution, meaning those strategies are evolving and that’s a good thing. Meaning companies like Twitter constantly are trying to learn about how these strategies are panning out. Right. Even within our teams, we ensure that continuously the review, what we do and how does it map to the strategy where it deviates and if it deviates, what is the underlying cause of the deviation? Right. And go back and try both to provide feedback to the upper management, as well as adjust how we are off.

Ryan: That makes sense. So much of machine learning happens under the hood. And, there’s a lot of people programming, different aspects of it and trying to create a cohesive. Approach it in the right way. So everybody’s clear as to exactly what it does. Our, I know we’re pretty notable challenges across projects I’ve been involved in. Do you feel like that’s something that you have to keep working with your teams on 

Foad: yes. Is the short answer. So with the Michelin project, The main challenge is the uncertainty around the project itself. Usually the problem statement is a high level and vague statement, right? Meaning, let’s say in the context of notification, you want to make sure that the recommended notifications are relevant to our customers. Right? This is a very vague statement. Definition of relevancy is definition of recommendation and that’s something that the team tries to do. Basically clarified by working with product managers, cross-functional partners, user research, and also experimenting with different ideas that you have, right. Even with that, the end result is uncertain. Meaning you don’t know that the model that you’re trying to build at the end of the day is actually going to improve the notion of relevancy or improve the experience of the users of Twitter. And the way that we deal with this is by now, this is part of the, oppressional culture, meaning that we know we are dealing with uncertain situations, right? And therefore, as we lay out our approaches, or as we lay out our roadmap, we try to create a healthy balance by some definition of health of a portfolio that is a mix of more certainty. And, projects that are higher risk or we don’t know exactly that experiment. 

Ryan: It’s interesting. I was just thinking about my own career. As you were talking about yours, where I’ve worked at a number of startups. I’ve also worked at a number of enterprise level organizations. And, I asked people who have been through the same thing, this question. Do you feel like the leadership quality in those two different environments?Are different. Like what, when you were starting Wanda, how you are a leader then versus how you are a leader now, working with a much bigger team with a product that’s impacting so many people, 

Foad: It is very different from my experience. for multiple reasons, one is with Wanda. The notion of leadership happened in the middle of our journey. Without any specific tipping point, meaning when you start a startup, you started with two or three people, right? It’s about just building a product, everybody working together gradually as things evolve gradually as the company grows, suddenly I realized, okay. So the notion of leadership is becoming An important part of my role in the company. Right. that came with a lot of different learning sports, painful learning as well as enjoyable ones. But the tutor is different, right? Meaning with Twitter, I started as an engineering manager, let’s say managing a team. We already were talking about the future, its implications and how we can get to where we are. Where we are is that I’m responsible for four different teams. the size of the say 40, 50 people, for instance. So there was more preparedness in terms of responsibilities. That’s come under the umbrella of leadership compared to the startup, which was my first.

Ryan: Yeah, that’s a good description of it. Yeah, I can see how it’s kind of a little bit more of a preset environment. But at the same time, like how you actually inspire people within that shift versus the startup environment. Let’s go back to your experience at Wanda for a second. There’s definitely a specific leadership style at the start of a startup, but when you get to the point of exiting a startup, how much did that shift? What kind of new things did you have to look at? What were some of the challenges at that point where you’re making that transition?

Foad: That’s a good question. So, first as you can imagine, Ryan, almost all experiences are different in startups, similar to what we had, but in the case of Wanda, when it started, the leadership aspect of my role was ensuring that. People are properly and rightfully excited about the mission of the company and the products that we are building, celebrating every moment of success within the company.And painting a clear picture as much as possible, about where we are headed as the company, the group. Different and new challenges that arguably are necessary, but not exciting. We faced and the leadership was. Shifted toward making decisions. And convene those decisions with the team and try to utilize their input and feedback for future decisions. Meaning down the road in Wanda, the team in general, not one or any startup or any company, the team looks for decisions to be made by the leader. Whereas at the beginning, it’s more collaborative. the execution mode and just like a 

Ryan: discovery process. Exactly. 

Foad: And I’m not thinking that this is the right way. It should be. I’m simply saying this was my experience. Right. I’m sure there are better ways to apply leadership skills at the beginning of a startup and towards the end. So that’s the biggest shift that I see. 

Ryan: I look at where you’re at now at Twitter, and it seems like your leadership is very much centered on solving a problem or solving a group of problems that are impactful in the world.

You’re similar to maybe the way you started Wanda. You came at a problem in a market, but as a company evolves and the leadership has to evolve within the evolution of the company, you kind of shift from just solving a problem to. Really leading people through change, right? I mean, that’s what it sounds like.

Foad: It is true. And let’s say right now on Twitter, if I want to say, what is the main area of my focus is it revolves around people, right? Meaning I don’t directly solve the notification problems. My focus is around the people who are actually solving those things and the various aspects that, people management, if you will, right? Because bullets at a startup environment or at a bigger company like Twitter at the end of the day, the biggest asset that the leader has is, is people That’s why I strongly believe that one of the primary focuses of a theater or a manager should be, their team, their people, and, their wellbeing across the board. Right. And how. The motivation to work with the common goals that the company has, if that is in place, which is a big, if it’s not trivial at all, I think the problems that need to be solved on the product and technical side will be much easier to deal with. 

Ryan: Yeah. That makes sense. Really, you are tied to the power of your team and how much you enable them. So, not to be funny here, but it is like in this particular arena, you know, training humans is a lot more complex than training machines , and really having to like, hear all of the complexities of the human experience within going through really heavy duty periods of innovation or, long nights of like getting the delivery done, things like that.

Foad: Yeah. I mean, training is, is one word to describe it, but in general, ensuring that, people, individuals on the team and the digitals that you work with. aligned with the objectives that you have. A lot of W’s of the company are motivated, have a feel safe in their working environment, right. Properly empowered and enabled. Right? All of these things are multidimensional complex problems to deal with. 

Ryan: Yeah. And sounds like you’re really getting help guide some people through big problems that face us all. And we really appreciate the work you’re doing there. Leadership is often about modeling. Good citizenship for an engineering leader, like Foad, someone that is at the foundation flow start restart. Taking two leadership roles is often about modeling good citizens. For an engineering leader, like Foad, someone that is at the foundational floor of training.AI, it’s even more important to shepherd a level of care for results of the products he creates, whether he was seeking to impact the healthcare industry or looking to help facilitate expansive social conversations at Twitter. He knows it’s vital to bring clarity and relevance to the results of a problem that is almost always uncertain.

Right. Welcome back everyone. We’re here with Fawad today to talk about his top five leadership tips. We’re excited to hear what he’s got. So let’s start with tip number one. What do you get for us? 

Foad: Empathy? And this is something that I’ve learned in more recent years of my experience, but at the end of the day, as we discussed before, people that you work with.People that you serve. I, our users are the biggest assets of what you do, and being able to have empathy toward them, understand where they’re coming from, understand their viewpoint. Isn’t extremely important. assets put a leader to make decisions and really understand where it is. 

Ryan: Yeah, to me, this is one of the most central leadership points I come across as well, because you’re right. You are discovering empathy for both the customer and empathy for the team. That’s collaborating to pull it off, and finding that balance. And it’s really important. All right, let’s jump to tip number two. What do you get? 

Foad: It is ROI where I have time. So we all are familiar with the ROI, which is return on investment, but here, the investment is time. At the end of the day. Your time is literally limited and knowing. What kind of return you’re getting when you’re spending time on a particular topic is very important. Otherwise detractors can easily spend hours or days or weeks on topics, which at the end of the day, the Delta between spending and days versus M days is very small. So I think it’s very important for leaders to really know how much time is worth spending on the topics that they have to do. 

Ryan: Yeah, it makes sense. What do you feel like this was a learned skill for you? Something you had to gain as you got further into your career, did you come in kind of knowing this? 

Foad: It was an, what helped me to learn?

It was feedback that I got from senior leadership, right? I mean, fortunately we have great mentors as you know, leaders at Twitter, and I’ve been given even explicit feedback about that . I should have spent more than a day on it. It means that you’re doing it wrong.

Ryan: yeah. it’s really easy to go in deep on something and forget that and, really thinking about the way you place the time, do you help guide your team and your team members to see the same thing? 

Foad: It’s one of those things that when I started to. Come over with it and learn to gradually get better at it. I found it very useful and that’s why I immediately started sharing it with people on my team. And even with my peers and colleagues, right, as this, as a concept that they should pay attention to, and also making observations on when I believe somebody is spending too much time on a topic and then having a discussion to see if maybe they can make a difference.

Ryan: That’s great. All right. Let’s jump to tip number three. 

Foad: It’s uncertainty and contradictions. I think this might sound like a self-contradictory statement, but something that is constant. And, leadership challenges are uncertainty and contradictions, meaning at the end of the day, a lot of problems that we try to solve. The problem statement is not fully defined and we are dealing with uncertain situations. And also we find ourselves even making decisions that might look contradictory to what we did. In a different setup, but I think that’s the reality of it. Meaning the dimensions of the problems that we try to solve at times gets so diverse that they may require decisions that may look contradictory, but it is the best way to deal with the uncertainties around these problems.

Ryan: Wow. This is a really good perspective. And I was relating to it a lot as you were talking through it, I also thought about contradictory elements, how your teams are perceived as well, because sometimes you have to make decisions where someone might only have a line of sight towards one aspect of it. And that might create uncertainty within those groups or within people who can’t see everything. So it really applies in a variety of ways. All right. Let’s jump to number four. 

Foad: And it’s a chronic discontent. I think for leaders, it is easy to end up focusing on what is in front of us and ensuring we do our best efforts to let’s say, execute, grow, and empower forks, but it’s important to us. Pay attention to chronic discontent and be aware of what is slowing us down, what can or should be changed and then go and try to make those changes. It is a slippery slope because making change at times can become very appealing. So the idea is not to go and find things to change, but rather be aware of things that are holding you or your team or your company’s bag and go and change them.

Ryan: What’s a way in which you communicate that discontent without being too negative. 

Foad: When I talk to my team or even to people that I’m hoping to join our teams, I present those as opportunities in the sense of being able to really understand what are the chronic situations that the team is operating under. That can be improved with a big impact. I think it’s an opportunity for folks to pay attention to and take those on. 

Ryan: Amazing. That’s really great. All right, let’s jump to your last tip. Tip number five. What do you got for us? 

Foad: This one is a trivial one, but I think it’s extremely important. And that has an impact at the end of the day. What we do has to be measured by the impact that we are making, whether that’s the product that we are driving, better efforts around hiring that you’re doing, whether it’s a company that you want to go and do fundraising for. Right. I think it’s extremely important for leaders to remember at the end of the day. Impact is what matters most and reviewing the impact that you’re making on a continuous basis and ensuring what you’re doing is going to actually lead or has the potential to lead to impact. I think it’s very critical. 

Ryan: Can you describe one or two times or different ways in which you’ve done that? 

Foad: let me give two examples, one in one, right? In one that let’s say in the middle of the journey that we had giving. And the startup and the excitement around doing a lot of things and our ability to move very fast. Because again, we are a startup. We always had a lot of interesting problems to go after and solve. But the way we try to choose which ones to go after the, did it based on the impact. Then it could affect our customers in that case being healthcare providers. And again, we can debate whether those decisions were right or wrong. I’m simply thinking of the story. But once we chose those based on impact, then it was very rewarding when we saw them at the same time.We did make decisions at Wanda, which was not based on impact, but it was based on how exciting the problem was, or how great it would be to solve. It’s very cool, which has significant negative implications. We actually did a direct to consumer application at Wanda targeting a women’s hearts. Which is very exciting, but we really didn’t understand what kind of impact we can make with this. And this was a failed effort and it led to a lot of challenges, at Twitter. We always set our goals based on the impact. That the projects are going to make both quantifiable impact as well as qualitative one. And then finally, again, it’s a different context, but, in places that I was not able to make an impact, right. , or the example that I briefly mentioned about that, the consumer about a women’s heart health. It was a big effort and a major failure. I mean, a magnitude that I can’t even begin to explain and faculty that came with a lot of painful learnings,

Ryan: Let’s keep going there for a second. How do you think the painful learnings translate back to you so that you can use them for the future? 

Foad: again, this is based on my own experience, but unfortunately the more painful the learning is. The easier the learning points can get right. In that particular case, some of the measure learning one was wrong, impact, which we covered. The second one, which was really painful. How you listen to people’s feedback and advice they give Z w I think everybody will agree that you need to have good mentors, advisors talk to people this and what they have to say. And arguably a lot of people do that, but then it actually manifests into a decision that you’re making, I think is a big one. Meaning back then, I used to talk to a lot of people. And, but then I looked back, I said, okay, I’m listening to what they’re saying? But at the end of the day, I am biasing my actions towards what I wanted to do. Right. And accept or dismiss people’s input. I think it’s very important to really put aside your own biases and try to see where the advice is coming from. What are the expertise and experiences behind their advice? And then really try to take them into account in your decision-making and not just listen to them. And this happened at about that consumer app. People told me don’t do it very explicitly. Now. I was like, yeah, they have a good point.

Right. That bot is very dangerous that you put at the end of the statement when you are thinking for yourself. 

Ryan: Yeah. And we’re told over and over again in the full color of business that I was holding a million times. No, but I still did it. It happened and that’s not always the case. It just doesn’t happen a lot of times, if you have a good trusted set of advisors, use shortcuts, huge mistakes by actually hearing them and taking them in.

And you ultimately have to make your own decisions and be your own, be responsible for what you want. Do. 

Foad: And this is, this goes back to the, in my opinion, the uncertainty, right? Meaning there are a lot of great stories about when people didn’t listen to overwhelming advice, that would point them in a different direction and they would really succeed.

 And those are great stories to learn from right. Again, it’s not like it’s a rule. That’s 10 people telling you not to do it. Don’t do it. But you can still do it, but do it after really putting your biases aside for a second. Yeah. And analyzing the situation and making it this.

Ryan: It is a huge part of holding that bright light out in front of everybody and making decisions as a leader. It’s central. And because whether or not people are telling you to do it, that’s a tough spot sometimes because you’re out in front.

Foad: Exactly.

Ryan: one thing Foad drove home over and over is that uncertainty is one of the only true constants that leaders face. And when you’re working in an environment that is constantly innovating at the edges of technology, it’s vital to lead with an eye toward the impact we are making on customers, team members and ultimately society, and the world of machine learning. The impact of the projects is an especially important consideration. Considering how it affects all of our lives and will ultimately affect the forthcoming generations.