Matteo Bettini
Research Scientist at Meta
Cambridge, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolMatteo Bettini is a research scientist at Meta and a Cambridge-based PhD in computer science focused on multi-agent and multi-robot systems, with six years of industry and academic experience. He blends rigorous research—studying heterogeneity and resilience in distributed agent systems—with hands-on ML engineering, having contributed to PyTorch’s reinforcement learning library by improving environment abstractions and parallel env handling. Matteo has interned on the PyTorch team and trained LLM agents for long-horizon tasks at Meta, demonstrating strength in both foundational algorithms and scalable engineering. As a guest lecturer and former supervisor at Cambridge, he translates complex topics into teaching and mentoring roles across robotics and programming languages. His background spans industry at AWS and Meta, top-tier research training, and active open-source work that eases integration and testing of RL algorithms. Colleagues would notice his combination of deep systems thinking and practical fixes that make research codebase-ready.
6 years of coding experience
4 years of employment as a software developer
Doctor of Philosophy - PhD Computer science (Multi-agent and multi-robot systems), Doctor of Philosophy - PhD Computer science (Multi-agent and multi-robot systems) at University of CambridgeBachelor of Engineering - BE Computer Engineering, Bachelor of Engineering - BE Computer Engineering at Politecnico di MilanoScientific High School Diploma, Scientific High School Diploma at Liceo Scientifico Statale “R. Donatelli – B. Pascal”
English, Italian
Github Skills (6)
pytorch10
machine-learning10
python10
reinforcement-learning10
multi-agent-reinforcement-learning9
testing8
Programming languages (7)
Github contributions (5)
pytorch/rl
Dec 2022 - Jan 2023
A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.
Role in this project:
ML Engineer
ML EngineerContributions:456 reviews, 14 commits, 128 PRs in 29 days
Contributions summary:Matteo contributed to the PyTorch RL library by addressing various aspects of the environment implementations. Their work involved refactoring and cleaning up base environment classes, as well as fixing bugs in batched environments. Furthermore, they made improvements to the `ParallelEnv` class concerning reset and done flags. The user also focused on the integration of a new library wrapper for the Vmas environment, allowing for easier integration and testing of RL algorithms within the library.
pytorchpythonreinforcement-learningprimitivedeep-reinforcement-learning
facebookresearch/BenchMARL
Oct 2023 - Mar 2025
A collection of MARL benchmarks based on TorchRL
Contributions:9 releases, 19 reviews, 181 PRs in 1 year 5 months
benchmarkmachine-learningmarlmulti-agentmulti-agent-reinforcement-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.