How Do I Break Into AI Product Management?

Watch the Episode

Transcript

Shobhit: Ready to move to the next level in your product career? I'm Shobhit from Intentional Product Manager. Join me as we discuss ways to help you stand out in your job search and your career so you can have more impact and make more money. Everyone, welcome to the Intentional Product Manager podcast. So with me is Saurabh Shrivastava, who's a seasoned product manager with a focus on defining long-term vision, highly differentiated products, awesome user experiences, establishing partnerships with other companies, big and small. He's currently a Group PM at Snap, working on many state-of-the-art AI/ML platforms — ML use cases like GenAI, which everyone's talking about, ads ranking, content ranking, user growth initiatives. Today we'll talk a lot about AI and machine learning based product management. So first of all, Saurabh, welcome to the podcast.

Saurabh: Thank you very much, thank you very much for having me here.

Shobhit: Saurabh, do you mind giving us a bird's eye view of your career journey?

Saurabh: Yeah, absolutely, would love to. So currently I'm working as a product manager at Snap, and specifically in that I'm working as a product manager for their machine learning platform — this is a platform that machine learning engineers at Snap use to create the state-of-the-art models which are used by our monetization teams to show the most relevant ads to you, or by our content team to show the most relevant stories or Spotlight to you, or maybe our growth team, for example, to show you the most recommended, like, best friend recommendation that you're most likely to chat with. So that's what I'm working on currently. But in my journey, I would say I've been very fortunate to work on different aspects of product management — I've worked as a product manager for very consumer-facing products, I've worked as the product manager for very enterprise-facing products, and I've worked as product manager for the platforms, the AI/ML platforms, that I'm working on for the last three, four years. So I'll maybe quickly dive into one of the examples of each, for people to understand what I'm referring to. I started my career as a consumer-facing product manager — this is when I was at Microsoft, working for their search engine Bing, and I worked on various scenarios on Bing. I think my first product management role was working on the Bing finance vertical — if you're searching for, say, Microsoft stock, and the experience that lights up is something I was working on. We used to work on toolbars back then, or like Edge browser's toolbar and things like that — so that's where I started, on the consumer-facing role. Fast forward several years, again in Microsoft with the Bing team, but this time in the Bing APIs team, where I worked as the enterprise product manager — we were working on different Bing APIs to power searches across the industry. A lot of people are actually not aware, but if you go to yahoo.com or duckduckgo.com, those are all results powered by Bing. So I was the PM for that team, and in my last role, lately, within Meta and Snap, I've been working as the platform PM — and both of these teams, actually, I've been working on their AI platforms.

Shobhit: So I'm going to ask you a lot about AI product management, but first you mentioned you've done these three, almost like people consider three verticals of product management — B2C, B2B, and then platform. When you were transitioning across these roles, did you have any sort of resistance, that hey, you're like a B2C person, now you're not B2B, or anything like that — and how did you overcome and position yourself for success in the next role?

Saurabh: Yeah, so for B2C to B2B it was not a very big problem for me, because I moved within Microsoft, so it was not that challenging — I think the thing that helped me the most there was my career track record, so I had a decent career track record working as a B2C PM, and people who were hiring me had access to my history of performance and everything, so that helped me a lot. Then from B2B to ML platform is where people see a lot of friction when they try to move, because there are different requirements from a consumer-facing product manager to say, the enterprise-facing product manager, in fact compared to platforms, because platforms are usually more horizontal, the projects are long-term compared to B2C, which is different. But I think a few things helped me a lot in this kind of movement when I did it from Microsoft to Meta — I would say, number one, at least for the interview, I had a very strong recommendation coming from one of my previous managers who had worked with me in the past, so that helped me at least get a foot in the door for the interview. And for the interviews, I had done a lot of preparation — I'm personally very interested in learning new things, so I had always been investing my energy in learning about AI/ML, working hard in that space, I've done a bunch of courses on Coursera and other places, so that gave me all the basic understanding of it, and I think I nailed the interviews pretty well — I knew the two most important interviews, like the technical rounds, and I feel I did a very good job at it, and I think that helped. And that helped me transition to platforms, and then Snap also started opening up more opportunities for me, and overall, whenever I've been carving out my career, I have been making sure that I don't get pigeonholed into one thing — so even when I was in Microsoft, within Bing, I had tiptoed a little bit on the AI side, or the platform side, and I used stories from that time to explain how I have learned in this space and contributed in this space. For example, for a brief period of time in Bing I worked on their relevance team, when I was just transitioning to AI, so that helped me a lot — I worked on their ML platform team as well for Bing, so that helped me a lot. Those stories helped me justify why I have this experience and why I would like it and can contribute towards it.

Shobhit: So was this move — now ML platform PM — was that like you planned it out, you were like, no, that's what I want to be, or is it sort of like happenstance, or both? Just eager to hear about what piqued your interest into this.

Saurabh: So in terms of how I moved to ML, it starts with when I was in Microsoft, working on the consumer-side B2C projects, back then, working in Bing — one thought that struck me is that if I'm working in Bing and I do not know relevance, there's no point working in this team — you're working in one of the surgeons of the company, one of the popular ones, and if you do not know how it works at the core, I won't be justifying my time here. So I consciously moved from a user-facing experience to the relevance team in Bing, and that was my first step in this direction. Then after working at Microsoft for a while, I wanted to move to Facebook, because I wanted to get an opportunity to work on very technical, AI-geeky stuff — I really loved my role at Microsoft, my last role at Microsoft, which was a wholesome product manager experience, but I wanted to do more of that, like I'm sure many of the PMs reaching out to you want to get into this field — I was in the same boat back then, I wanted to get into this role a little bit more, and Facebook was, and still is, kind of leading in the space, so I wanted to be at the right place there.

Shobhit: One more question on that — you said you had to be in the relevance team, why?

Saurabh: Yes, so the primary reason was, if I'm in Bing and I've spent like five years and do not understand how the core of the product works, I felt I'm not justifying my presence here — so that's why. It was not for career ambition, it just felt right to know the core of a product that I am giving five years to, or maybe more.

Shobhit: Got it, got it, no, totally, that makes complete sense. Let's talk more about — you mentioned how you got into Meta — like, anything else that helped you differentiate yourself from other candidates, other than what you've already mentioned here?

Saurabh: I think at different points of time, when I've done a career move, there have been different factors that have played a role in getting me that role. I can talk about when I moved to the Bing team in Microsoft, when I moved from India to the US, and then we can quickly talk about Microsoft to Facebook as well. So when I wanted to move from India to the US, I think a few things helped — the move to the relevance team helped, because it opened up more opportunities in the headquarters here for me, and in Bing there were a lot of openings on the relevance role. Second thing, I actually researched about the team I was applying for, and I found out what their problems are right now, and I wrote a small, like, a two-pager kind of thing of what that team should be doing, and I shared that with the hiring manager — I think that impressed him, so he was on a business trip to India, and after I shared this doc with him, when he came, he met with me, and we spent like an hour discussing that doc. So I think this whole thing, that he saw that I have the drive and I have the technical competency, so he sort of started batting for me internally — he's a great guy and everything, so I think that helped me a lot, and he's a great guy, I'm still good friends with him right now. So that was something which helped me a lot moving from India to the US. And from within Microsoft to Facebook, I think what helped me the most was a very thorough preparation for the interview — so, it's basically a two-step process, right, first getting an interview call, and second, how do you do in the interview. So to get the interview call, you have to have the right resume, and you have to have the right recommendation — to get the right resume, I had done the right projects, but I wanted to show that I'm very interested in the space, and I basically wanted to learn that space, so I spent a lot of time doing a bunch of courses — if you see on my LinkedIn, I've done Coursera and a few more things out there, just before I started applying for this role, because I wanted to be super comfortable talking in the language that the team is talking in. So that helped, and of course the recommendation helped, from the hiring manager. So that was for the interview call — and during the interview, I think it's very important to research about the company you're applying for, and understand what the interview structure is, what the pattern is, like at Facebook at that time, there were four or five categories, I'd done very thorough preparation for how to nail those interviews. So in some of them, when I went in, I knew what they were going to ask, I gave them the answer, and there it is — so, product sense and those categories, but the other specific ML interviews as well. I started my career at Facebook as a TPM, it was more technical rounds back then — they had a round in which they give you a technical problem to solve, and all these courses I'm talking about had helped me a lot in answering those questions, and second, they also ask about, tell me about some project you've worked on in the past — you have to be able to start from what the product is and go as deep as you can technically, do not just be at the technical layer, you have to bridge the whole gap. And I think that helped — so I think these are the two interviews in which I did decently well, and I personally feel that interviews are a little weighted towards one or two interviews — once you nail down the most important ones, it becomes easier to get the other ones too.

Shobhit: Let's talk more about — you mentioned you did a bunch of courses — which ones helped, and what really helped, was it like you putting on your resume that you've done that course, or something else that was really valuable from that?

Saurabh: So what usually helps when you do these courses is if you get the opportunity to talk to the team — because once you talk to the team, they understand that you know the space. The interview structure in different companies is very different — like if you go to say Google, for example, it doesn't matter if you've done any courses or not, because their interview structure is very different, and if you've not gotten the audience with the hiring manager, you do not even get that opportunity. So it helps only when you get the opportunity to talk to the hiring manager — then you can talk in a language and express that this is why I'm valuable in this space. That's one. And secondly, I feel, sometimes in some of the interviews, depending on the company, these help — in case of Facebook, the technical round was very much pivoted on machine learning, which I was applying for — I think one of the questions they asked me was how would you build a classifier for some scenario, and then how would you scale it, and all those things. Since I knew this space well, I could talk about all the core details, like loss function, this and that, so that helped me a lot — I was able to explain them the basic concept, then take it up a notch by talking about how do you scale it to the Facebook level and all those things. But if you do not know the basic fundamentals, which you get from these courses, you will just be talking the paper version of it — and that's what I've seen when I've been taking interviews, that people learn certain, like, keywords, when you talk about say scale or distributed systems, and they just throw a bunch of these in there, but they do not have the fundamental concepts clear. I think the courses usually help you get that confidence — it won't directly help, but it indirectly helps you a lot.

Shobhit: What I'm hearing is, like, courses were a way of you really understanding the space, what's going on, what the technical terms mean, and now you're a lot more conversational in your discussions, rather than throwing certain terms around. Now that you've been on the other side, and you've been an AI platform product manager for a while, so what's changing — what's changing in product management careers with AI being rapidly adopted?

Saurabh: I think one thing to understand for everyone is that, like we talked about, three verticals, right — B2C, B2B, and platform — within platform there is AI platform. Likewise, there's one more way of dissecting this thing — like, AI product managers can be divided into two major categories, one is applied ML product manager, and second is AI platform product manager. Applied ML are the ones who use machine learning to show you the best, most relevant content — like TikTok recs, everyone is chasing applied ML there. And then this ML platform product manager is the one who builds the platform that these engineers can use to do this thing. So when people are applying for these product manager roles, they should be very conscious of which one they are more interested in, and whether they're applying to the right one or not — for example, applied ML is closer to the customer, and ML platform is closer to the tech. So people have to be careful about that.

Shobhit: Let's talk about both — like, how does somebody decide which one they should go for?

Saurabh: It depends what you have in mind — if you want to be closer to the customer, it makes sense to be the applied ML product manager, because, for example, I'll take an example from Snap — if you want to show the most relevant Spotlight, or most relevant friend request to the user, you should be in applied ML. Now if you want to understand the tech space very well — because all these applied ML PMs are using a platform, for you to contribute and add value, you have to understand this whole space — if you want to build that competency, be in the ML platform space.

Shobhit: And then let me ask a variation of this — how should they look at it in terms of what's most suited to their experience? Are there certain characteristics that would make them more suitable for an applied AI role versus an AI platform role?

Saurabh: I think one thing is from the resume itself — if you've worked on something in the past, and you have a rough idea, that will help. Second, one good way is to look at the companies out there and see the type of role they are offering — for example, if you want to be the ML platform guy, Snowflake and Databricks are the kinds of companies, do you see yourself working at companies like this? If you want to be the applied ML guy, you might be working in the ads org of Facebook, or say TikTok, or Reels — do you want to be there? So these are some very high-level thumb rules that can help you get a directional idea of do I want to go in this direction or this direction.

Shobhit: Next question is, okay, you've been through this transition — if somebody came to you and said, hey, I'm a strong PM, I've hardly ever had any exposure to AI, what should I do to get into this space, what advice would you give them?

Saurabh: Yeah, so one thing I would recommend for folks is, it depends what stage of your career you're in — if you're early in your career, people are more open to give you that opportunity, give you that break, even if you have not — if you're like a senior PM or principal PM, the chances are lower. But if you still want to break into the space, the best way is, within your company, find opportunities where you can do more of this ML stuff — just get familiar with this space. It helps you, first, in understanding do you enjoy the space, because the space is very different — if you've been a UX-heavy PM, you will not be deciding, hey, should this be blue or this be green, should I move five pixels here or not — you'll be working more closely with engineers here, so the space is very different — do you enjoy the space or not, that is one opportunity for you to understand. And it is also an opportunity for others to see that you can add value, and within your company people are usually more open to give you those opportunities — this is what I did, at least in Microsoft — just find your sister team that you've been working with, figure out how you can move there, so that helps a lot. It's all about getting that break, and after that there's a snowball effect — you can start learning more, you can start contributing more, you beef up your resume with that — but that one break helps, and usually people closer to you are readier to give you that break.

Shobhit: When you are hiring PMs at Snap or at Meta, and they're applying for AI/ML roles, what are maybe some of the common failure patterns that you see — that, you know, they didn't do a good job, and so they didn't land the role in the AI/ML space?

Saurabh: I think the most common pattern I have seen is that people are not aware about the space — they have not spent time in the technical domain, or they do not know how they can solve the customer problem, and those are the big red flags for me. Many times, generalist PMs come and assume that they will be able to contribute in the space, but it's a little more challenging, because the terminology, the lingo, is different, the people you are building product for are different, and you have to figure out how do you add value — for example, I'll take this — you are building product for engineers, with the help of your engineers — both know more than you, right, in the technical space — how do you add value? You have to bring some niche to the table that both of these guys might not be able to do — this is very tough, because you're playing on their turf, this is fully technical — so you do all sorts of, like, competitive study or things like that, but you have to figure out how would you add value here.

Shobhit: I love that point of view — I used to work in dev tools, and so my engineers were building products for other engineers, and went through the same challenges. So then let's look a little forward — do you see any major changes in product management, the job market, over the next year, and how would you tell people to navigate those?

Saurabh: I think, in the last year and a half, the job market has been very brutal — I'm sure you must have seen it as well — but I have seen that it's opening up a little bit now, we have seen openings coming up, and, like, we have a bunch of openings in Snap now, we've seen a bunch of openings coming up in other companies too — this has started opening up. I definitely see there is a bias towards applied ML kind of PM roles right now — there are many companies hiring for those roles, or at least the platform roles — I'm noticing that there are good things coming up in that space.

Shobhit: Anything that PMs should do to take advantage of the market opening up, and especially more hiring in AI/ML specific roles?

Saurabh: I think while switching companies, it's very important to understand the company you're applying for, especially their interview patterns — they're very different, for example, Amazon versus say Google or Facebook. Amazon is always looking in the rearview mirror, like, tell me about your past, where you have shown competence on these leadership principles — that's how they judge people, so it's more about the STAR framework, and telling your stories the right way. Facebook and Google tend to be more on the creative side — hey, can you do this, I have this business problem for you, how would you solve this thing. So understand that, and do a lot of mocks and practice for those, because the interview process is very noisy — those people are looking for very specific things, and if you say it in those 45 minutes, it works, otherwise you're out.

Shobhit: Saurabh, thank you so much for taking time to be on this podcast and sharing your expertise on product management, especially AI/ML.

Saurabh: Perfect, thank you very much, thank you for having me here — I've been following your posts on LinkedIn, they've been very insightful and very engaging as well, so I love it, I appreciate it.

Shobhit: Last thing I always ask — like, how do people find you, should they connect with you on LinkedIn, or do you write somewhere?

Saurabh: I think LinkedIn is the best way to reach out to me. Thank you very much, bye-bye.

Shobhit: Hey, be sure to check out our website at intentionalproductmanager.com to see how you can level up in your career.