# Fermi Ma

![Fermi Ma](/resources/cognition/portraits/fermi-ma.webp)

NYU Courant Asst. Prof. (quantum crypto) | Researcher at Cognition on SWE-1.7 & FrontierCode  
NYU Courant Institute of Mathematical Sciences  
United States

## At Cognition

- Role: Researcher (Research)
- Joined: Not public
- How they joined: Directly
- Before Cognition: Assistant Professor of Computer Science, NYU Courant Institute of Mathematical Sciences
- Overlapped before Cognition with other Cognition people at: MIT; Princeton University (https://www.autumn.ai/resources/cognition/worked-together)

## Identity
- LinkedIn: https://www.linkedin.com/in/fermi-ma-70a8b543
- X: @fermi_ma
- Website: https://fermima.com/
- Domain: cognition.com

## Notes
- Credited in Research on Cognition's SWE-1.7 launch (2026-07-08), the RL-post-trained coding model built on a Kimi K2.7 base and trained inside the Devin harness; author list reads '... Fermi Ma, Allen Liu, Yang Liu, Sampriti Panda, Jacob Teo, Ray Wang, Gary Chang, Steven Cao, Silas Alberti'. (https://cognition.com/blog/swe-1-7)
- His public notes on OpenAI's Unitary Synthesis Problem solution state: 'All of the following text was written by me, with extensive help from Devin, an AI agent that had access to a PDF generated by Astra.' (https://fermima.com/notes/usp-notes.pdf)
- Listed in the Research credits for 'Introducing FrontierCode', Cognition's coding benchmark with rubric-based grading, alongside Eric Lu, Ben Pan, Deniz Birlikci, Sam Lee, Ray Wang, Rohan Choudhury, TC Qin, Carlo Baronio and Silas Alberti. (https://cognition.com/blog/frontier-code)
- Listed third in Research credits for FrontierCode 1.1 ('Eric Lu, Ben Pan, Fermi Ma, Alex Lombardi, ...'), the benchmark version SWE-1.7 reports its 42.3% headline score on. (https://cognition.com/blog/frontier-code-1.1)
- In those notes he describes his workflow verbatim: 'I would write a bullet, send it to Devin, and ask if it was the correct next step. If I got something wrong, I'd ask Devin to explain the issue, and then I'd rewrite the bullet and check again.' (https://fermima.com/notes/usp-notes.pdf)
- Assistant Professor of Computer Science at NYU's Courant Institute; research focuses on quantum computing and its interplay with cryptography, complexity theory, and physics. (https://fermima.com/)
- Co-authored with Hsin-Yuan (Robert) Huang 'How to Construct Random Unitaries' (STOC 2025, QIP 2025 plenary), proving pseudorandom unitaries exist, a question open since 2017. Quanta Magazine covered it. (https://fermima.com/)
- Teaching CSCI-GA 3033 Quantum Cryptography at NYU Courant in Fall 2026. (https://fermima.com/)
- On X he follows several Cognition researchers, including Deniz Birlikci, Ben Pan, Carlo Baronio, Jeffrey Ling, and Sushant Sachdeva (U. Toronto faculty who is also a Cognition researcher), plus Cheng-Yuan (Sam) Lee, whose bio reads 'Research @cognition | prev intern @windsurf | 2x ICPC World...'. The FrontierCode research group he co-authors with includes ex-Windsurf and ICPC people. That he follows these researchers is a lead, not proof of any team assignment. (https://x.com/fermi_ma/following)
- Recent X activity has shifted toward LLMs. On 2026-02-06 he wrote: 'This is a very neat result: given a dataset and a target system prompt like "reply in Spanish," they show you can select a subset of the data such that fine-tuning an LLM on that subset causes the model to behave *as if* it were given that system prompt!' (https://x.com/fermi_ma/status/2019899862297497883)
- Cognition, his AI employer, signed a definitive agreement to acquire Windsurf, the agentic IDE (IP, product, trademark), in July 2025, days after Google DeepMind hired Windsurf's CEO. The SWE-1.x model line he contributes to began as Windsurf's in-house models. No source shows he personally came from Windsurf. (https://cognition.com/blog/windsurf)
- Announced on X on 2025-06-05: 'This fall, I'll be joining NYU as an assistant professor of computer science. The department is betting big on quantum computing, and I'm thrilled to be part of it.' (https://x.com/fermi_ma/status/1930685303842635997)
- PhD in Computer Science from Princeton (2015-2020), advised by Mark Zhandry; Zhandry's group page lists him as 'Fermi Ma. → Postdoc at UC Berkeley → Asst Prof at NYU'. (https://mzhandry.github.io/)
- His Cognition benchmark work is already cited in academic ML research: the 2026 arXiv paper RECLAIM cites 'Eric Lu, Ben Pan, Deniz Birlikci, Sam Lee, Ray Wang, Rohan Choudhury, Fermi Ma, TC Qin, Carlo Baronio, and Silas Alberti' (FrontierCode). (https://arxiv.org/html/2609.28850v1)
- On X (2024-10-15): 'Can efficient quantum circuits behave like truly random unitaries to any efficient observer? The existence of such circuits, called pseudorandom unitaries (PRUs), has been open since 2017.' (https://x.com/fermi_ma/status/1846206369038467362)
- Simons-Berkeley Postdoctoral Fellow at UC Berkeley (2021-2025) hosted by Umesh Vazirani; taught CS294 'Cryptography in a Quantum World' at Berkeley in Spring 2023. (https://fermima.com/)
- Previous roles include Postdoctoral Researcher at University of California, Berkeley; Graduate Student at Princeton University. (https://www.linkedin.com/in/fermi-ma-70a8b543)
- Co-authored a CRYPTO 2022 paper with James Bartusek, Yael Kalai, Alex Lombardi, Giulio Malavolta, Vinod Vaikuntanathan, Thomas Vidick and Lisa Yang. Lombardi, a later Cognition co-author, is a long-running collaborator going back at least to then. (https://people.csail.mit.edu/vinodv/pubs.html)
- Previous roles include Postdoctoral Researcher at University of California, Berkeley; Graduate Student at Princeton University. (https://www.linkedin.com/in/fermi-ma-70a8b543)
- Location is United States. (https://www.linkedin.com/in/fermi-ma-70a8b543)
- Location is United States. (https://www.linkedin.com/in/fermi-ma-70a8b543)

## Experience
- Assistant Professor of Computer Science at NYU Courant Institute of Mathematical Sciences (Jan 2026 - Present)
- Postdoctoral Researcher at University of California, Berkeley - Berkeley, California, United States (Sep 2021 - Dec 2025)
- Graduate Student at Princeton University - Princeton, New Jersey (Sep 2015 - Aug 2021)
- Assistant Professor of Computer Science at NYU Courant Institute (Jan 2025 - Present)
- Researcher (SWE-1.7, FrontierCode) at Cognition

## Education
- Princeton University - Doctor of Philosophy (Ph.D.), Computer Science - (Jan 2015 - Jan 2020)
- Massachusetts Institute of Technology - Bachelor's Degree, Course 18: Mathematics - (Jan 2011 - Jan 2015)

## Sources
- https://www.linkedin.com/in/fermi-ma-70a8b543
- https://fermima.com/
- https://cognition.com/blog/swe-1-7
- https://x.com/fermi_ma
- https://x.com/paddingtonbear
- https://people.csail.mit.edu/vinodv/pubs.html
- https://arxiv.org/html/2609.28850v1
- https://cognition.com/blog/frontier-code
- https://cognition.com/blog/frontier-code-1.1
- https://fermima.com/notes/usp-notes.pdf
- https://www.jstage.jst.go.jp/article/ipsjjip/23/3/23_293/_pdf/-char/en
- https://ntt-research.com/dr-mark-zhandry-post-quantum-cryptography-and-more/
- https://mzhandry.github.io/
- https://cs.nyu.edu/dynamic/about/news/colloquium/1378/
- https://simons.berkeley.edu/people/fermi-ma
- https://cognition.com/blog/windsurf
- https://techcrunch.com/2025/07/14/cognition-maker-of-the-ai-coding-agent-devin-acquires-windsurf/
- https://www.linkedin.com/posts/john-preskill-154abab3_quanta-magazine-reports-on-the-beautiful-activity-7312164853542961152-WAno
- https://dblp.org/pid/146/0144
- https://x.com/fermi_ma/following
- https://x.com/fermi_ma/status/2019899862297497883
- https://x.com/fermi_ma/status/1930685303842635997
- https://x.com/fermi_ma/status/1846206369038467362


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