Inside AI Product Management: From Infrastructure to Innovation

Introduction to AI in Product Management

Rina Alexin | 00:02–00:54
Hi everyone and welcome to Productside Stories, the podcast where we reveal the very real and raw lessons learned from product leaders and thinkers all over the world. I’m your host, Rina Alexin, CEO of Productside. And today I’m joined by Digvijay Singh, the Senior Product Manager at YouTube. Digvijay is working on a team using AI and machine learning models to optimize the ad experience. He also has a data science background and has held product manager roles at Swift Financial, later acquired by PayPal, and Google. In this episode, we’ll be unpacking the evolving role of AI in product management and its implications on the product manager career. Welcome, Digvijay. Excited to have you.

Digvijay Singh | 00:54–00:58
Thank you, Rina, for the introduction. I’m glad to be here.

Rina Alexin | 00:58–01:09
I always like to start with your story—that’s why we named this series Productside Stories. How did you end up in product management?

Digvijay Singh | 01:09–03:30
I started my career in engineering undergrad in India. After that, I got into data science consulting, where I advised clients in the credit card and insurance industries about machine learning models. Then I went to a startup called Swift Financial, where I built machine learning models for digital loans for small businesses. Once I was done with that, I wanted to bring together my data science skills with business knowledge. I got my MBA at Columbia Business School and moved into product management at Google. I’ve been primarily working on AI and ML products—first in the Google Play Store, and now in YouTube Ads, focusing on ad optimization and user experience.

The Evolution of AI and Its Impact on PM Roles

Rina Alexin | 03:30–03:46
So for everyone who’s not a premium user—are you the person helping advertisers know when the clicks happen, how to convert better? Is that what you’re working on?

Digvijay Singh | 03:46–04:42
Great question. If I had a penny for everyone who starts with that question about premium, I’d be in a different income bracket. The way most ad systems work is through demand-side and supply-side signals. I work on the supply side—optimizing how many ads a user should see and tailoring experiences based on whether someone is ad-sensitive or not. We try to ensure the AI models balance user experience with advertiser value.

Rina Alexin | 04:42–05:42
That’s fascinating. And this is a great start because, as you said, AI has actually been in our lives for a long time, hasn’t it?

Digvijay Singh | 05:42–06:12
Yes, absolutely. I’ve been at Google since 2019 and in data science before that. Neural networks, deep learning—all these terms have been around for years. What’s new today is the scale and accessibility of large language models. These systems have revolutionized how we think about AI, but the foundation has existed for over a decade.

Understanding Different AI Product Management Roles

Rina Alexin | 06:12–06:30
So how do you see AI reshaping PM roles?

Digvijay Singh | 06:30–08:47
I usually divide AI product management into three categories: infrastructure, model, and application. Infrastructure PMs work on data centers and compute resources—the backbone of AI. Model PMs focus on developing foundational models like Gemini or GPT, optimizing metrics and algorithms. Application PMs apply those models to solve real-world problems—things like coding assistants, recommendation engines, or ad optimization. The roles are interconnected but require different skill sets, from technical understanding to product intuition.

The Application Layer: Unique Challenges and Opportunities

Rina Alexin | 08:47–09:01
The majority of PMs probably work in the application layer, right? What’s unique about those roles?

Digvijay Singh | 09:01–11:32
Yes, exactly. Application-layer PMs focus on making AI useful and practical. Unlike traditional apps, AI products evolve with data. It’s like managing an intern—one that can generate creative solutions but sometimes make mistakes. You need to guide it through clear goals and evaluation frameworks. We don’t just ship code; we train behavior. That’s what makes it so dynamic and challenging.

Evaluating AI Models: The Role of Evals

Rina Alexin | 11:32–11:45
You mentioned evaluation frameworks. Can you explain what evals are and why they matter?

Digvijay Singh | 11:45–14:03
Evals are essentially the success metrics for AI systems. They tell the model what “good” looks like. Think of it like teaching an intern—you define the goal, provide examples, and then measure performance. Evals give structure to AI behavior, reducing hallucinations and making systems predictable. They act like OKRs for AI, ensuring alignment between user intent and model output.

The Model Layer: Innovations and Skill Requirements

Rina Alexin | 14:03–14:15
Let’s talk about model-layer PMs. What makes those roles special?

Digvijay Singh | 14:15–16:43
Model-layer PMs work closer to the innovation core. They define how foundational models evolve—what new use cases to pursue and what metrics define success. It’s a lot of zero-to-one work: exploring possibilities like text-to-video generation or multi-modal intelligence. These PMs need to combine strategic thinking, technical fluency, and creativity to shape what’s next.

Future Trends and AI PM Skills

Rina Alexin | 16:43–17:00
What trends do you see shaping the future of AI PM roles?

Digvijay Singh | 17:00–19:09
We’re still early in the AI era—similar to how the internet evolved in the 90s. At first, value accumulated at the infrastructure level—companies like Cisco built the foundation. Over time, applications dominated. Right now, infrastructure and model-layer PMs are in high demand, but soon hybrid PMs—those who understand both tech and customer—will be the most valuable.

Advice for Aspiring AI Product Managers

Rina Alexin | 19:09–19:29
So for PMs wanting to break into AI—what advice would you give?

Digvijay Singh | 19:29–36:21
Think of AI like mentoring an intern. Define goals clearly and evaluate thoughtfully. Learn how to write strong evals—they’re your compass. Also, play with AI tools every day. Build intuition through experimentation. The more you engage, the sharper your product sense becomes. Don’t just read about AI—experience it. That’s how you build confidence and credibility as an AI PM.

Closing Thoughts

Rina Alexin | 36:21–40:33
Thank you so much for joining me, Digvijay. For our listeners, if you found today’s conversation valuable, please share it and subscribe to Productside Stories so you never miss an episode. Keep learning, keep experimenting, and keep innovating. I’m Rina Alexin—thanks for listening.

Digvijay Singh | 39:59–End
Thank you, Rina. It’s been a pleasure. Feel free to reach out to me on LinkedIn if you’d like to connect or discuss more about AI and product management.