Vedant Misra Engineer | Scientist | Advisor | Investor | Entrepreneur Vedant Misra's work focuses on advancing the state of the art in AI and using it to build new products. As a founding member and technical lead on Google DeepMind's Gemini team, he has driven reasoning capabilities in science, math, programming, and long-horizon agentic tasks in Gemini 1, 2, and 3. He led Minerva, achieving a breakthrough 50% accuracy on the MATH benchmark, and worked on the state-of-the-art PaLM and PaLM 2 models. Previously, he worked with Ilya Sutskever to lead the Reasoning, Algorithms, and Multimodal teams at OpenAI, where they released Codex, which launched the modern wave of AI code generation. Before that, he founded Kemvi (acquired by HubSpot) and led the Machine Learning Labs team at HubSpot, developing new products using large neural networks with text, images, and audio. He studied physics, mathematics, and economics at Columbia University as a Rabi Scholar and worked on black hole research at the Institute for Strings, Cosmology, and Astroparticle Physics. His work has been covered by TechCrunch, Fortune, Wired, Technology Review, and others, with publications and patents spanning machine learning, natural language processing, program synthesis, medical imaging, human-computer interaction, black hole physics, and quantitative finance.