Michele Catasta
President & Head of AI · Replit
President and Head of AI at Replit, having joined as VP of AI in 2023 and been elevated in 2024, the year he architected and led the launch of Replit Agent — a release the company credits with more than two orders of magnitude of revenue growth. An EPFL computer science PhD, he was Head of Applied Research at Google X and Google Labs working on large language models including PaLM and PaLM 2, and earlier a research scientist at Stanford pioneering Transformer architectures applied to source code.
Last reviewed: Sep 18, 2026
Michele Catasta
President & Head of AI Replit
Last Reviewed: September 18, 2026
Executive Summary
Michele Catasta is President and Head of AI at Replit, the browser-based software development platform. He joined as VP of AI in 2023 and was elevated in 2024, the year he architected and led the launch of Replit Agent — the product that turned Replit from a coding environment into a system that writes and deploys software from a natural language description, and which the company credits with more than two orders of magnitude of revenue growth. He is unusually well prepared for that specific problem: he holds a computer science PhD from EPFL, led Applied Research at Google X and Google Labs working on PaLM and PaLM 2, and before that was a research scientist at Stanford pioneering the application of Transformer architectures to source code.
Career Highlights
- Architected and led the 2024 launch of Replit Agent, credited by the company with more than two orders of magnitude of revenue growth.
- Joined Replit as VP of AI in 2023 and became President and Head of AI in 2024.
- Former Head of Applied Research at Google X and Google Labs, contributing to PaLM and PaLM 2.
- Pioneered applying Transformer architectures to source code as a research scientist and instructor at Stanford.
- Held research positions at the MIT Media Lab, Google, and Yahoo Labs.
- Speaks frequently on applied AI and coding agents, including at Web Summit, SaaStr, and the ASU GSV Summit.
Professional Journey
Catasta took his PhD in Computer Science at EPFL in Switzerland and built an academic and industrial research career across several of the institutions where machine learning moved from theory into deployment: Yahoo Labs, the MIT Media Lab, and Google. At Stanford he worked as a research scientist and instructor, pioneering the application of Transformer architectures to source code — treating programs as a language a model could learn to read and write. That line of research is the direct antecedent of every AI coding assistant now in use, and he was working on it before the commercial category existed.
He went on to lead Applied Research at Google X and Google Labs, focusing on large language models and contributing to PaLM and PaLM 2, Google’s foundation models of that period. “Applied research” is the operative phrase: the work concerned getting capability out of models and into things people could actually use, which is a different discipline from advancing the state of the art.
He joined Replit as Vice President of AI in 2023 and was elevated to President and Head of AI in 2024. Replit’s platform runs entirely in the browser, removing the setup burden that keeps many people from programming at all, and that architecture turned out to be unusually well suited to what came next. In 2024 Catasta architected and led the launch of Replit Agent, which takes a description of a desired application and builds, runs and deploys it — feasible in Replit’s case precisely because the environment, execution, and hosting already lived in one place. The company credits the release with more than two orders of magnitude of revenue growth.
As President and Head of AI he oversees AI strategy and research direction, and speaks widely on coding agents and applied AI at events including Web Summit, SaaStr, the ASU GSV Summit, and The AI Conference.
Education
École Polytechnique Fédérale de Lausanne (EPFL)
PhD, Computer Science
