Carlo Baronio
Carlo Baronio
Math @ Stanford; Researcher / Engineer at Cognition working on RL and coding-agent systems (Devin, Kevin, SWE-grep).
Projects
Ideas brought to life. A look at the products, companies, and creative works built from the ground up.
Kevin: Multi-Turn RL for Generating CUDA Kernels
Research project and paper demonstrating multi-turn reinforcement learning to generate and iteratively refine CUDA kernels (Kevin-32B). Co-authored with Pietro Marsella, Ben Pan, Silas Alberti and others; demonstrated strong results on kernel benchmarks.
Kevin-32B — Cognition blog / technical report
Cognition blog post describing the Kevin-32B model and multi-turn RL approach for CUDA kernel generation, authored by Carlo Baronio and collaborators.
SWE-grep / SWE-grep-mini
Research and engineering work on fast agentic retrieval models (SWE-grep family) to improve context retrieval speed and parallelism for coding agents; Cognition blog co-authored by Baronio.
Devin (Cognition product updates & research contributions)
Contributions and activity on Cognition's coding-agent product 'Devin' including product updates and research-driven improvements to agent capabilities.
Parallel Sampling via Autospeculation (theory / systems work)
Co-authored work accepted to academic venues (listed in STOC/NeurIPS materials) on parallel sampling and autospeculation; aligns with research on efficient sampling strategies.
In the Spotlight
A collection of articles, interviews, and viral content where achievements have been shared.
Recognitions
Formal validation from the top. A showcase of official awards, grants, and government acknowledgements.
Bronze Medal — International Mathematical Olympiad (IMO)
July 2023Bronze medalist at the International Mathematical Olympiad (listed on LinkedIn profile).
Gold Medal — Italian Mathematical Olympiad
May 2023Gold medal at the national Italian Mathematical Olympiad (listed on LinkedIn).
Bank of Italy Prize for the Study of Mathematics and Informatics
2020Recognition from the Bank of Italy for achievements in mathematics and informatics (listed on LinkedIn).
Work Experience
Key roles that defined the journey.
Researcher / Engineer
2024-presentCo-authored Cognition research and contributed to engineering efforts on coding agents (Devin), multi-turn RL (Kevin), and fast retrieval models (SWE-grep). Sources: Cognition blog posts and LinkedIn activity.
Undergraduate / Student (Mathematics)
2023-09 - 2027-07 (expected)Mathematics student at Stanford University (LinkedIn education entry).
Education
Academic foundation and learning journey.
Undergraduate (Mathematics)
Socials
Connect and explore.