- Joined
- Sep 2026
- How they joined
- Directly
- Before Cognition
- Vector InstituteFaculty Member
- Status
- Current
- On 2025-11-21 he announced on X, verbatim: "I believe we're at the doorstep of a revolution in how we'll do theoretical research and this is just the beginning! A good time to announce that I've taken leave from UofT to be a part of the revolution on the inside at @OpenAI! :)"; OpenAI researchers DJ Strouse, Jason Lee and Dan Roberts replied with welcomes.Nov 21, 2025I believe we're at the doorstep of a revolution in how we'll do theoretical research and this is just the beginning! A good time to announce that I've taken leave from UofT to be a part of the revolution on the inside at @OpenAI! :)
- Sushant Sachdeva announced that he has taken leave from UofT to join OpenAI.Nov 2025I believe we're at the doorstep of a revolution in how we'll do theoretical research and this is just the beginning! A good time to announce that I've taken leave from UofT to be a part of the revolution on the inside at @OpenAI! :)
- Sushant Sachdeva has been a Member of Technical Staff at OpenAI since 2025.2025
- Sushant Sachdeva is on leave at OpenAI starting Fall 2025.2025
- Sushant Sachdeva was covered in a 2022 Quanta Magazine article, 'Researchers Achieve Absurdly Fast Algorithm for Network Flow', and on the Quanta Science Podcast (Apple, Google, Spotify).2022
- Sushant Sachdeva received a Google Faculty Research Award, 2017 (USD 43,335).2017
- Joshua Wang was a mentee of Sushant Sachdeva in Summer 2017 and is a Research Scientist at Google.2017
- Sushant Sachdeva gave a talk on 'Fast Algorithms for Optimization and Learning on Graphs' at Google, New York, in May 2016.May 2016
- Qiuyi (Richard) Zhang was a mentee of Sushant Sachdeva in Fall 2016 and is a Software Engineer at Google.2016
- Sushant Sachdeva was a Visiting Faculty Researcher at Google from 2024 to 2025.2024-2025
- Sushant Sachdeva was a Research Scientist at Google from 2016 to 2017.2016-2017
- Sushant Sachdeva worked at OpenAI.Can't believe it's been 18 years. Sangram Raje, Ajit Nema The world tomorrow will be very different, but being able to think through problems deeply will be the difference that will make your work stand out.
RL training science
Distillation for large language models (llms)
Visiting scientist at the Fast Algorithms program.
I believe we're at the doorstep of a revolution in how we'll do theoretical research and this is just the beginning! A good time to announce that I've taken leave from UofT to be a part of the revolution on the inside at @OpenAI! :)
Can't believe it's been 18 years. Sangram Raje, Ajit Nema The world tomorrow will be very different, but being able to think through problems deeply will be the difference that will make your work stand out.
💯 This is pretty certainly the future of tcs/maths. We should prepare for it as communities and clearly articulate/incentivize the kind of works our venues want to promote.
You are right to push back! It's not masterful prose, just a word salad.
every day I feel like shouting into the void: if you wish to be known as a professional of high taste, do not post AI slop the game is long, and you risk your reputation sinking rapidly amongst the people who matter. Thank you for your attention to this matter!
"Load bearing" https://t.co/gFaKFr9ILX
Not my kink to take abusive emails pretending to be funny @dbrand https://t.co/V7HWYIZEai
Very accessible, brief and enjoyable accounts on the status of the millennium problems by one of the 🐐
YC being like, why wait, we'll create our own singularity
FWIW, 5.6 and Fable are definitely generating mathematical results that are publishable at top venues. However, my view (and I believe of most academics) is that the write-ups and exposition are horrendous and definitely unacceptable at any top venue @ajeya_cotra
iTs NoT a PuRe LLM unLeSS yoU ImPleMeNt it with ExCEl SpReAdSheeTs
@aryehazan @OpenAI Glad I could help out with this! Maybe the models find measure theory counter-intuitive too :D j/k
