Deepak Tiwari came up through engineering and strategy consulting at Accenture before moving into product, and his career has run through some of the most consequential AI and infrastructure builds of the last 15 years, across Google, Lyft, Turing, and Meta. At Turing, he served as CPO and built the training data platform used by OpenAI, Anthropic, DeepMind, and Meta. Before that, he helped take Google Cloud from fewer than $0 to $500M in annual revenue. He's now at Meta, leading product across ranking, relevance, and generative AI systems at Instagram scale.

In this episode, we cover:

• What it took to pitch Google Cloud's CEO on a net-new product from scratch, and what happened when customers started using it in ways the team never anticipated
• How the enterprise transformation playbook that worked at Google Cloud became the blueprint for Turing's 10x growth
• What working directly with the frontier labs taught him about why data is the real moat in AI
• Building Meta's first generative AI ads product with no prior playbook, and reading transformer and diffusion model papers from scratch to understand what was actually possible
• Why the biggest drivers of user experience at Lyft had nothing to do with the app, and everything to do with backend ML
• The three levers that actually improve a model, and why good PMs need to hold all three in view
• Why the manager who barely had time for one-on-ones taught Deepak the most important career lesson

An interesting conversation for PMs and product leaders who want to understand how AI products actually get built, from the infrastructure up.