# Rohan Choudhury

![Rohan Choudhury](/resources/cognition/portraits/rohan-choudhury.webp)

![Rohan Choudhury 2](/resources/cognition/portraits/rohan-choudhury.webp)
![Rohan Choudhury 3](/resources/cognition/portraits/rohan-choudhury.webp)

Researcher at Cognition training SWE-1.6/1.7 coding models for Devin and Windsurf; CMU Robotics PhD  
Cognition  
San Francisco, California, United States

## At Cognition

- Role: Member of Technical Staff (Engineering)
- Joined: Dec 2025
- How they joined: Directly
- Before Cognition: Student Researcher, ByteDance
- Overlapped before Cognition with other Cognition people at: Nuro; Caltech (https://www.autumn.ai/resources/cognition/worked-together)

## Identity
- LinkedIn: https://www.linkedin.com/in/rohan-choudhury
- X: @rchoudhury997
- GitHub: https://github.com/rccchoudhury
- Website: https://rccchoudhury.github.io/
- Domain: cognition.com

## Notes
- Third author on 'SWE-1.7: Frontier Intelligence at a Fraction of the Cost' (2026-07-08), after co-leads Ben Pan* and Carlo Baronio*; SWE-1.7 came from RL pipeline improvements on top of Kimi K2.7, trained in Devin's harness. (https://cognition.com/blog/swe-1-7)
- On SWE-1.7 launch day he wrote on X: "when I joined Cognition, training our own frontier-class model felt like a dream that was far away. We got there in 6 months - faster than even our most optimistic projection. With this team, truly anything feels possible. Try SWE-1.7 with Cerebras today - you'll be blown away." (https://x.com/rchoudhury997/status/2074883916423631224)
- Lead (first-listed) author on Cognition's 'Introducing SWE 1.6: Improving Model UX' (2026-04-07), the model optimized for intelligence and model UX, shipped generally available in Windsurf at up to 950 tok/s with a 3-month free tier. (https://cognition.com/blog/swe-1-6)
- Announced on X on 2026-03-01: "I guess now is as good a time as any to announce that I finished my PhD and started as a researcher at @cognition! This has been a wild first project with an absolutely insane team, excited for what's coming next"; Cognition colleagues Ben Pan (@ybenpan) and Timothy Luong replied. (https://x.com/rchoudhury997/status/2028224671586615323)
- Quoted by Forbes in Aug 2024 as a CMU robotics PhD using Codeium's AI coding assistant, adopted 'on his former colleagues' suggestion': "Most of the time it just works out of the box." Codeium (Windsurf) was later acquired by Cognition. (https://oodaloop.com/briefs/news-briefs/this-ai-coding-engine-can-process-100-million-lines-of-code-at-once/)
- On FrontierCode 2.0 he wrote: "With FC2.0 we've tried to measure the models on "true" software engineering skills, and the result is a ranking that imo matches with the vibes. it turns out this isn't quite saturated yet (for now)" (https://x.com/rchoudhury997/status/2115139953789374859)
- Co-author of Cognition's FrontierCode code-quality benchmark (2026-06) and FrontierCode 1.1 (2026-07-07) with Eric Lu, Ben Pan, Fermi Ma, Alex Lombardi, Deniz Birlikci and others; he tweeted: "Agents are writing more code than ever, yet we all viscerally know the code isn't high-quality. We worked hard to make a benchmark to quantify this "slop" aspect. We expect this benchmark will saturated sooner than later but models that perform well on it will be a joy to use." (https://cognition.com/blog/frontier-code-1.1)
- On Devin Desktop he wrote: "a year ago i was hand-typing print statements and ipdb.set_trace(). today that feels quaint. devin desktop is a big step towards the future of software!" (https://x.com/rchoudhury997/status/2062031306322817128)
- Co-author of 'An Early Preview of SWE-1.6 and Research Update' (2026-03) with Carlo Baronio, Ben Pan, Sam Lee, Eric Lu, Steven Cao, Adam Zweiger and others; the preview landed around the time he publicly announced joining, matching his 'wild first project' tweet. (https://cognition.com/blog/swe-1-6-preview)
- Cognition acquired Windsurf (formerly Codeium/Exafunction), taking its IP, product, trademark and brand in July 2025, days after Google DeepMind hired Windsurf's CEO Varun Mohan; SWE-1.x models he trains ship in Windsurf (Devin Desktop). (https://cognition.com/blog/windsurf)
- Windsurf/Codeium co-founder Varun Mohan previously worked at Nuro, where Rohan was a software engineer 2019-2021. (https://www.frederick.ai/blog/varun-mohan-windsurf)
- Reshared Cognition's funding announcements on LinkedIn: "proud to be a part of this!" on the $1B+ raise at a $26B valuation (2026-05-27) and "so lucky to be a part of this ride!" on the $2B+ raise at a $48B valuation (2026-09-09). (https://www.linkedin.com/posts/rohan-choudhury_cognition-has-raised-over-2b-at-a-48b-valuation-activity-7503462675725651971-ripn)
- On Devin Fusion (hybrid-model harness) he wrote: "fusing the power of multiple models intelligently is really hard but can yield superior performance with much better cost. proud of our team for taking the lead on this! https://t.co/VGLgsVbNT9" (https://x.com/rchoudhury997/status/2071662717145096515, https://t.co/VGLgsVbNT9)
- 2025 preprint 'Accelerating Vision Transformers with Adaptive Patch Sizes' (APT) with JungEun Kim, Jinhyung Park, Eunho Yang, Jeni and Kitani; code at github.com/rccchoudhury/apt. (https://rccchoudhury.github.io/)
- PhD at CMU Robotics Institute advised by Kris Kitani and Laszlo Jeni, focused on making visual models more efficient at understanding and generating visual content; supported by the NSF GRFP Fellowship. (https://rccchoudhury.github.io/)
- Student Researcher at ByteDance Seed from 2025-03, collaborating with Peter Lin (Shanchuan Lin) and Lu Jiang on accelerating video generation. (https://rccchoudhury.github.io/)
- Open source: official RLT implementation github.com/rccchoudhury/rlt (NeurIPS 2024), which cuts wall-clock fine-tuning time for video transformers by about 30% while matching baseline performance; plus github.com/rccchoudhury/apt for adaptive patch size ViTs. His GitHub account lists 18 public repositories. (https://github.com/rccchoudhury/rlt)
- First-author NeurIPS 2024 spotlight (top 3%) 'Don't Look Twice: Faster Video Transformers with Run-Length Tokenization' (RLT) with Guanglei Zhu, Sihan Liu, Koichiro Niinuma (Fujitsu), Kitani and Jeni; code at github.com/rccchoudhury/rlt. (https://rccchoudhury.github.io/)
- Public headline says "Researcher @ Cognition". (https://www.linkedin.com/in/rohan-choudhury)
- PhD at CMU Robotics Institute advised by Kris Kitani and Laszlo Jeni, focused on making visual models more efficient at understanding and generating visual content; supported by the NSF GRFP Fellowship. (https://rccchoudhury.github.io/)
- Interned at Meta FAIR working with Jing Huang on efficient video understanding. (https://rccchoudhury.github.io/)
- ECCV 2024 paper 'Video Question Answering with Procedural Programs' (ProViQ) with Koichiro Niinuma, Kitani and Jeni; Niinuma is at Fujitsu Research, matching his 2023 Fujitsu Research internship. (https://rccchoudhury.github.io/)
- Before the PhD was a software engineer at Nuro developing trajectory forecasting models for self-driving vehicles (2019-2021). (https://rccchoudhury.github.io/)
- At Caltech (B.S. CS, 2015-2019) explored multi-agent reinforcement learning with Yisong Yue. (https://rccchoudhury.github.io/)
- Named an Outstanding Reviewer for CVPR 2025. (https://rccchoudhury.github.io/)
- Previous roles include Student Researcher at ByteDance; Research Scientist Intern at Meta; Research Scientist Intern at Fujitsu Research; Software Engineer at Nuro. (https://www.linkedin.com/in/rohan-choudhury)
- Location is San Francisco, California, United States. (https://www.linkedin.com/in/rohan-choudhury)

## Experience
- Member of Technical Staff at Cognition - San Francisco, California, United States (Dec 2025 - Present)
- Student Researcher at ByteDance - New York, New York (Feb 2025 - Nov 2025): Seed - Video Generation
- Research Scientist Intern at Meta - Menlo Park, CA (May 2024 - Sep 2024): Meta FAIR, Video Understanding
- Research Scientist Intern at Fujitsu Research - Pittsburgh, PA (May 2023 - Aug 2023)
- Software Engineer at Nuro - Mountain View, California (Aug 2019 - Aug 2021): Behavior ML

## Education
- Carnegie Mellon University - Ph.D, Robotics - (Jan 2021 - Jan 2026)
- Caltech - B.S, Computer Science - (Jan 2015 - Jan 2019)

## Sources
- https://www.linkedin.com/in/rohan-choudhury
- https://cognition.com/blog/swe-1-6
- https://cognition.com/blog/swe-1-7
- https://x.com/rchoudhury997
- https://rccchoudhury.github.io/
- https://x.com/rchoudhury997/status/2028224671586615323
- https://oodaloop.com/briefs/news-briefs/this-ai-coding-engine-can-process-100-million-lines-of-code-at-once/
- https://cognition.com/blog/swe-1-6-preview
- https://cognition.com/blog/frontier-code-1.1
- https://cognition.com/blog/windsurf
- https://www.frederick.ai/blog/varun-mohan-windsurf
- https://github.com/rccchoudhury/rlt
- https://campuspubs.library.caltech.edu/3173/01/Issue%208%2C%202016-17.pdf
- https://www.linkedin.com/in/rohan-choudhury?miniProfileUrn=urn%3Ali%3Afsd_profile%3AACoAAAzMOrMBZdDFN--IqxeW-qhYQP3L6KCHVZw
- https://github.com/rccchoudhury
- https://cognition.com/blog/frontier-code
- https://x.com/rchoudhury997/status/2074883916423631224
- https://x.com/rchoudhury997/status/2115139953789374859
- https://x.com/rchoudhury997/status/2062031306322817128
- https://www.linkedin.com/posts/rohan-choudhury_cognition-has-raised-over-2b-at-a-48b-valuation-activity-7503462675725651971-ripn
- https://x.com/rchoudhury997/status/2071662717145096515
- https://t.co/VGLgsVbNT9


Researched with Autumn (https://www.autumn.ai) from public sources. Updated 2026-11-04. Profile: https://www.autumn.ai/resources/cognition/people/rohan-choudhury · Everyone at Cognition: https://www.autumn.ai/resources/cognition/people