Prafulla Dhariwal

Prafulla Susil Dhariwal is an Indian-born computer scientist and a Technical Fellow at OpenAI in San Francisco, where he has worked since 2016 (research intern, then research scientist, Head of Multimodal, Technical Fellow) and co-created models and systems including GPT-3, DALL-E 2, and GPT-4o. He represented India at the International Mathematical Olympiad in 2011 (honourable mention) and 2012 (gold). He holds a Bachelors in Computer Science and Mathematics from MIT (2013-2017). His industry internships were at Pinterest (software engineering, summer 2014) and The D. E. Shaw Group (quantitative analyst, summer 2015).

Notes4

• Quant employment verification: CONFIRMED. The D. E. Shaw Group is a quantitative investment firm (systematic/algorithmic trading and hedge fund, deshaw.com); the subject held a Quantitative Analyst Intern role there May 2015 to Aug 2015 in New York, per his LinkedIn experience section, building a model predicting cancellation activity on US stock exchanges and a trading strategy correlating cancellation activity with market P&L. Identity is supported because the same LinkedIn profile carries his MIT education, OpenAI roles, and IMO/Physics-olympiad honors matching the IMO 2012 participant record. This is a single internship and does not indicate later quant employment.May 2015
• Co-created and led major OpenAI generative-model programs: GPT-3, DALL-E 2 (lead author on the diffusion paper), Consistency Models, Point-E, and led the multimodal team behind GPT-4o.2024
• Represented India at the International Mathematical Olympiad twice: honourable mention in 2011 (score 282, 50.1%) and gold medal in 2012 in Argentina (score 33, 94.2%), per the official IMO participant record.2012
• Sole quantitative-finance line on his record is the summer-2015 D. E. Shaw internship; all other roles are academic research or software/AI, so a finance recruiting narrative should not be read into his trajectory.2024
• Personal site lists continuing generative-model publications including DALL-E 3, Consistency Models, and Improved Techniques for Training Consistency Models, showing diffusion and image generation were his focus through 2023-2024.2023
• Won a gold medal at the International Physics Olympiad in addition to the IMO medals, and was selected in India's top 30 for the International Olympiad in Informatics Training Camp (IOITC) 2013.
• Undergraduate research at MIT spanned two labs: Torralba's lab at CSAIL (2015, deep-learning depth estimation) and Prof. Poggio's lab at the Center for Brains, Minds and Machines (2016, DNN/CNN/RNN models for invariant speech representations), filling both summers he was not in industry.2016
• Scored 295 in Physics/Chemistry/Math in Class XII (2013) and 330/360 in JEE-Mains, earning the Abasaheb Naravne Memorial Award from the Maharashtra state board.May 17, 2024