Log inSign up
Ashvin Nair
Cursor
101 posts
user avatar
Ashvin Nair
Cursor
@ashvinair
RL foundations @cursor_ai. Prev: o1, o3, Code Interpreter @openai, 9 years learning to poke by poking at UC Berkeley
Berkeley, CA
ashvin.me
Joined December 2014
2,858
Following
2,931
Followers
  • Pinned
    user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Oct 29, 2025
    Some exciting new to share - I joined Cursor! We just shipped a model 🐆 It's really good - try it out! cursor.com/blog/composer I left OpenAI after 3 years there and moved to Cursor a few weeks ago. After working on RL for my whole career, it was incredible to see RL come alive
    user avatar
    Cursor
    @cursor_ai
    Oct 29, 2025
    Introducing Cursor 2.0. Our first coding model and the best way to code with agents.
    Image
    00:00
    Image
    Composer: Building a fast frontier model with RL · Cursor
    From cursor.com
    221K
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Oct 22, 2021
    A PyTorch re-implementation is now available - see github.com/rail-berkeley/… Use IQL for all your offline RL and online finetuning needs!
    user avatar
    Ilya Kostrikov
    @ikostrikov
    Oct 13, 2021
    Excited to present our work with @ashvinair and @svlevine, Offline RL with Implicit Q-Learning (IQL), a simple method that achieves SOTA performance on D4RL arxiv.org/abs/2110.06169 and works 4x faster than prior SOTA github.com/ikostrikov/imp… Thread below
    Image
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Sep 10, 2020
    Blog post on accelerating online RL with offline data: bair.berkeley.edu/blog/2020/09/1… w/ code. AWAC is great for learning online with prior data (demos, off-policy data, etc). Joint with @mihdalal @abhishekunique7 @svlevine
    Image
    GIF
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Aug 10, 2021
    Replying to @RyanDavidReece
    Took a minute to understand but its this effect causing: thestar.com/news/insight/2… For designing cockpits, out of 4,063 pilots, not a single airman fit within the average range on all 10 dimensions.
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Aug 8, 2024
    Headed to Amherst for @RL_Conference - I assume I'll find everyone organically but drop me a message if you'd like to catch up!
    2.3K
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    May 30, 2024
    A little shook at the extent that ChatGPT memory distilled me into my essence
    Image
    2.3K
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Jun 3, 2021
    Super excited to release latest work on visuomotor affordance learning (VAL)!! Highlights: 1. VAL utilizes a large prior dataset to fine-tune in new scenes with just 5 minutes of online robot interaction 2. Our high-quality robot interaction dataset now at
    Age-restricted adult content. This content might not be appropriate for people under 18 years old. To view this media, you’ll need to log in to X. Learn more
    sites.google.com
    sites.google.com
    Problem Statement How can robots learn about affordances from prior datasets and, when faced with new and unfamiliar environments, utilize this knowledge to practice relevant skills and update their...
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Sep 12, 2024
    Extremely excited about this release and for this next phase of progress with reinforcement learning+LLMs!
    user avatar
    OpenAI
    @OpenAI
    Sep 12, 2024
    We're releasing a preview of OpenAI o1—a new series of AI models designed to spend more time thinking before they respond. These models can reason through complex tasks and solve harder problems than previous models in science, coding, and math. openai.com/index/introduc…
    2.4K
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Nov 6, 2025
    Replying to @gabriel1
    Mamdani is certainly more abundance-pilled than either Cuomo or Silva (check out his podcast with Derek Thompson, he might not agree with every point but Mamdani seems to understand the issues pretty deeply); who were you hoping for?
    1.5K
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    May 5, 2024
    In Vienna for ICLR - reach out if you’re around and want to meet up!
    1.6K
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Mar 25, 2025
    Image
    885
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Oct 30, 2024
    Impressive!
    user avatar
    Murtaza Dalal
    @mihdalal
    Oct 30, 2024
    Can my robot cook my food, rearrange my dresser, tidy my messy table and do so much more without ANY demos or real-world training data? Introducing ManipGen: A generalist agent for manipulation that can solve long-horizon robotics tasks entirely zero shot, from text input! 1/N
    Image
    00:00
    1.3K
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Oct 13, 2021
    Really excited about new work with @ikostrikov and @svlevine on offline RL. Implicit Q-Learning (IQL) gets great results offline and finetuning further online, and is really fast and simple to implement. See Ilya's thread for details
    user avatar
    Ilya Kostrikov
    @ikostrikov
    Oct 13, 2021
    Excited to present our work with @ashvinair and @svlevine, Offline RL with Implicit Q-Learning (IQL), a simple method that achieves SOTA performance on D4RL arxiv.org/abs/2110.06169 and works 4x faster than prior SOTA github.com/ikostrikov/imp… Thread below
    Image
  • user avatar
    Ashvin Nair
    Cursor
    @ashvinair
    Mar 23, 2023
    Super excited for ChatGPT plugins - a way for LLMs to interact with the external world
    user avatar
    Greg Brockman
    OpenAI
    @gdb
    Mar 23, 2023
    We’ve added initial support for ChatGPT plugins — a protocol for developers to build tools for ChatGPT, with safety as a core design principle. Deploying iteratively (starting with a small number of users & developers) to learn from contact with reality: openai.com/blog/chatgpt-p…
    Image
    00:00
    2.1K

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
Advertisement
Advertisement