# Carlo Baronio

Research Lead at Cognition; co-lead of SWE-1.7, first author of Kevin-32B; IMO 2023 bronze medalist  
Cognition  
Stanford, California, United States

## At Cognition

- Role: Research Lead (Research)
- Joined: Apr 2025
- How they joined: Directly
- Before Cognition: Undergraduate Researcher @ Hazy Research, Stanford Artificial Intelligence Laboratory (SAIL)
- Olympiads: IMO bronze 2023 (https://www.autumn.ai/resources/olympiads/carlo-baronio-olp-4bd91ec0838758d08f2e)
- Overlapped before Cognition with other Cognition people at: Stanford University (https://www.autumn.ai/resources/cognition/worked-together)

## Identity
- LinkedIn: https://www.linkedin.com/in/carlo-baronio
- X: @carlobaronio
- Domain: cognition.ai

## Notes
- Promoted to Research Lead at Cognition in October 2026 after ~18 months as a Research Engineer (from April 2025). (https://www.linkedin.com/in/carlo-baronio)
- Co-author of SWE-2 (Sept 2026); posted verbatim: "SWE-2 is our first attempt to scale intelligence/cost and on FrontierCode it matches Fable 5.1 at 64% lower cost. We achieved this by developing a penalty that preserves the shape of the test-time scaling curve during training.". (https://x.com/carlobaronio/status/2098087016697389512)
- Co-lead (equal-contribution author, listed second after Ben Pan) of SWE-1.7, Cognition's July 2026 model shipped in Windsurf that the blog says 'reaches frontier-level intelligence at a much lower cost'. (https://cognition.com/blog/swe-1-7)
- First author (equal contribution with Pietro Marsella and Ben Pan, advised by Silas Alberti) of Kevin-32B, the first open model trained with multi-turn RL to write CUDA kernels, built as a Stanford-Cognition collaboration. (https://cognition.com/blog/kevin-32b)
- Posted verbatim on SWE-1.7 launch: "It's a good model sir?!! Pls check out our blog post there are a lot of cool findings/stuff we do to get huge gains from just RL on high quality data at scale. There is no wall 🧱" (https://x.com/carlobaronio/status/2074898372126027947)
- Co-author of SWE-1.6 (April 2026), released generally available in Windsurf, the IDE Cognition acquired in July 2025; his model work therefore ships directly into the acquired Windsurf product (Devin Desktop/Windsurf line). (https://cognition.com/blog/swe-1-6)
- Co-author of SWE-grep and SWE-grep-mini (Oct 2025), fast RL-trained multi-turn context-retrieval models powering Windsurf's Fast Context subagent. (https://cognition.com/blog/swe-grep)
- Current role is Research Lead at Cognition. (https://www.linkedin.com/in/carlo-baronio)
- Co-author of FrontierCode (June 2026), Cognition's hard coding eval, and listed in Research acknowledgments of FrontierCode 1.1 (July 2026), which became the benchmark SWE-2 claims are measured on. (https://cognition.com/blog/frontier-code)
- Posted verbatim on SWE-1.6 preview: "Nice milestone for our research effort: SWE-1.6 (early preview) is a meaningful step up from 1.5 without sacrificing speed (950 tok/s). Strong SWE-Bench Pro results, and this is just the beginning!" (https://x.com/carlobaronio/status/2028228687666315296)
- Kevin paper (with Simon Guo of Stanford/KernelBench added as co-author) was accepted at ICLR 2026; Google Scholar shows ~117 citations for it. (https://scholar.google.com/citations?user=DwvvTgQAAAAJ&hl=en)
- Featured speaker at CoreWeave Fully Connected 2026, presenting on training Cognition's SWE models with asynchronous RL at global scale; CoreWeave's bio says he works on algorithms and training infrastructure spanning Kevin-32B, SWE-grep and Cognition's SWE models. (https://www.coreweave.com/resources/videos/training-cognition-swe-2-asynchronous-rl-at-global-scale)
- Cognition CEO Scott Wu publicly praised Kevin-32B as 'really great work' from @carlobaronio, @pmmarsella and @ybenpan; Carlo replied verbatim: "Thanks Scott! More to come soon 😊" (https://x.com/ScottWu46/status/1919949534446354774)
- Kevin-32B weights are open-sourced on Hugging Face under cognition-ai/Kevin-32B, his main public open-source artifact. (https://huggingface.co/cognition-ai/Kevin-32B)
- Before Cognition: Undergraduate Researcher at Stanford's Hazy Research lab (SAIL), Dec 2024-Jun 2025, working with Benjamin Spector (ThunderKittens kernel lineage). (https://www.linkedin.com/in/carlo-baronio)
- Co-leads (with Silas Alberti) a Systems Reading Group with Cognition on Oct 5 covering async RL/pipeline RL, delta weight updates and sparsity. (https://sfsrg.org/)
- Summer 2024 research in Stanford's CS Theory Group with Prof. Nima Anari on parallel sampling, producing 'Parallel Sampling via Autospeculation' (STOC 2026) with Anari, CJ Chen, Alireza Haqi, Frederic Koehler, Anqi Li and Thuy-Duong Vuong. (https://alirezahaqi.github.io/)
- Bronze medal at IMO 2023 (Chiba, Japan) for Italy with 22 points (P1 7, P2 6, P4 7, P5 2), rank 220. (https://www.imo-official.com/participant_r.aspx?id=33732)
- Italy's IMO 2023 team: ITA1 Carlo Baronio, Simone Foti, Pietro Gualdi, Luca Sartori, Denis Tusca, Enrico Zonta; all six medalled and Italy placed 17th of 112. (https://www.oliforum.it/viewtopic.php?t=22639)
- Public headline says "Research @ Cognition". (https://www.linkedin.com/in/carlo-baronio)
- Bronze medal at the Balkan Mathematical Olympiad 2023 (Antalya) with 25 points; teammates included Edoardo Balistri, Raffaele Botticella, Simone Foti, Andrea Nanni and Denis Tusca; Italy placed 9th of 22. (https://olimpiadi.dm.unibo.it/2023/06/07/balkan-mathematical-olympiad-2023-5-bronzi-per-litalia/)
- I am an undergraduate student at Stanford University, interested in the intersection of Mathematics, Physics and AI; this comes from the profile summary and preserves the person's own positioning. (https://www.linkedin.com/in/carlo-baronio)
- Previous roles include Research Lead at Cognition; Research Engineer at Cognition; Undergraduate Researcher @ Hazy Research at Stanford Artificial Intelligence Laboratory (SAIL); Research @ CS Theory Group at Stanford University Department of Computer Science; Research @ DET at Politecnico di Torino. (https://www.linkedin.com/in/carlo-baronio)
- X bio reads 'member of quantized staff @cognition'; account created Jan 2024 with ~414 followers, posting rarely and mostly reposting Cognition research launches. (https://x.com/carlobaronio)
- Location is Stanford, California, United States. (https://www.linkedin.com/in/carlo-baronio)

## Experience
- Research Lead at Cognition (Oct 2026 - Present)
- Research Engineer at Cognition (Apr 2025 - Oct 2026)
- Undergraduate Researcher @ Hazy Research at Stanford Artificial Intelligence Laboratory (SAIL) (Dec 2024 - Jun 2025): Working with Benjamin Spector.
- Research @ CS Theory Group at Stanford University Department of Computer Science (Jun 2024 - Sep 2024): Working with Prof. Nima Anari on parallel sampling algorithms.
- Research @ DET at Politecnico di Torino (Apr 2020 - Dec 2022): Working with Prof. Francesco P. Andriulli on computational electromagnetics.

## Education
- Stanford University - Mathematics - (Jan 2023 - Dec 2025)

## Sources
- https://www.linkedin.com/in/carlo-baronio
- https://cognition.com/blog/swe-1-7
- https://cognition.com/blog/kevin-32b
- https://www.imo-official.com/participant_r.aspx?id=33732
- https://www.coreweave.com/resources/videos/training-cognition-swe-2-asynchronous-rl-at-global-scale
- https://x.com/carlobaronio
- https://www.oliforum.it/viewtopic.php?t=22639
- https://huggingface.co/cognition-ai/Kevin-32B
- https://x.com/ScottWu46/status/1919949534446354774
- https://cognition.com/blog/swe-1-6
- https://sfsrg.org/
- https://alirezahaqi.github.io/
- https://scholar.google.com/citations?user=DwvvTgQAAAAJ&hl=en
- https://github.com/abhinav30219/math_reasoner
- https://cognition.com/blog/frontier-code
- https://cognition.com/blog/swe-grep
- https://www.polito.it/ateneo/comunicazione-e-ufficio-stampa/poliflash/early-research-honors-school-tra-i-nuovi-studenti-il
- https://olimpiadi.dm.unibo.it/2023/06/07/balkan-mathematical-olympiad-2023-5-bronzi-per-litalia/
- https://cognition.com/blog/swe-2
- https://cognition.com/blog/frontier-code-1.1
- https://x.com/carlobaronio/status/2098087016697389512
- https://x.com/carlobaronio/status/2074898372126027947
- https://x.com/carlobaronio/status/2028228687666315296


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