Bansal, a graduate of IIT Kanpur, holds dual degrees in mathematics and statistics and later earned his PhD in computer science from the University of Massachusetts Amherst, specialising in meta-learning, deep learning, and natural language processing. His career trajectory includes a mix of academic rigour and hands-on experience at the world's biggest tech companies.
 started in 2012 with a stint as an analyst at Accenture Management Consulting in Gurugram. He then spent over two years at the Indian Institute of Science (IISc) in Bangalore as a research assistant, focusing on Bayesian modelling and inference methods.
Over the years, he interned at Facebook, Google, Microsoft, and OpenAI in roles involving deep learning for natural language processing, knowledge graphs, and reinforcement learning. At UMass Amherst, he also served as a graduate research assistant, where he delved into deep learning for NLP.
In 2022, Bansal joined OpenAI full-time as a member of technical staff. There, he worked alongside co-founder Ilya Sutskever on reinforcement learning (RL) for reasoning, co-creating the organisation's first AI reasoning model, o1. His contributions are widely seen as foundational to the development of ChatGPT's internal reasoning abilities.
His new home,  was formally launched by Mark Zuckerberg on Monday. The division is tasked with developing artificial general intelligence (AGI), and is being co-led by former Scale AI CEO Alexandr Wang and ex-GitHub CEO Nat Friedman. According to Bloomberg and The Wall Street Journal, Meta is aggressively hiring elite AI researchers from OpenAI, Anthropic, and DeepMind, sometimes over four years. While the specifics of Bansal's deal remain unclear, reports speculate his offer could include stock-based compensation and massive compute access..
The scale of these offers has caused tension in the industry. OpenAI CEO Sam Altman recently said Meta attempted to lure his staff with although he claimed "none of our best people" had accepted. But Bansal's exit, along with , suggests otherwise. Other researchers who have reportedly joined Meta include Lucas Beyer, Alexander Kolesnikov, Xiaohua Zhai, Jack Rae, and Johan Schalkwyk.
With Meta currently lacking a publicly available AI reasoning model, Bansal's appointment could help expedite the development of it. His expertise in building systems that can learn, reason, and plan is expected to shape Meta's next-generation LLMs and possibly rival OpenAI's o3 or DeepSeek's R1.
While Meta spokesperson Andy Stone has downplayed claims of exaggerated salaries, insiders believe Bansal's hiring is part of a broader strategy to fast-track AGI research.