Lucas Schott
Research Engineer Deep Reinforcement Learning (DRL) And Robustness Of Autonomous Agents at IRT SystemX
Paris, Ile-de-France
Summary
👤
Senior🎓
Top SchoolLucas Schott is a research engineer and PhD candidate based in Paris with nine years of experience specializing in deep reinforcement learning, robustness, and adversarial methods for autonomous systems. At IRT SystemX and Sorbonne Université he has driven projects on robust DRL agents for automotive, maritime and rail simulators, combining adversarial training, ensemble methods and hybrid planning to meet safety requirements. He contributed to the popular HighwayEnv simulator by refining kinematic observations—work that directly improves how DRL agents perceive traffic scenarios. His research blends theoretical rigor with practical deployment, including verification/hybridization methods for RAG systems and defenses against adversarial attacks on LLMs. He also teaches probabilistic and ML foundations at ENSAE and Sorbonne, bringing academic depth to applied engineering challenges.
9 years of coding experience
1 year of employment as a software developer
Bachelor's degree, Computer Science, Bachelor's degree, Computer Science at University of StrasbourgPhD, Robustness of Deep Reinforcement Learning Algorithms, PhD, Robustness of Deep Reinforcement Learning Algorithms at Sorbonne University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (7)
gymnasium10
environ10
python10
autonomous-driving10
reinforcement-learning10
enviroment10
pandas8
Programming languages (1)
Github contributions (5)
Farama-Foundation/HighwayEnv
May 2020 - Oct 2022
A minimalist environment for decision-making in autonomous driving
Role in this project:
Back-end Developer
Back-end DeveloperContributions:10 commits, 2 PRs, 8 comments in 2 years 5 months
Contributions summary:Lucas primarily focused on modifying the `highway_env` environment, specifically the kinematic observation system. Their contributions involved adding, modifying, and removing parameters related to vehicle observations. They addressed bugs related to observation logic and the `close_vehicles_to` function, refining the observation process and fixing conditional statements. These changes directly impacted how the environment perceives and processes information about vehicles, which is crucial for reinforcement learning agents.
pytorchdecision-makingautonomousdecisionminimalist
lucasschott/highway-env
May 2020 - Oct 2022
An environment for autonomous driving decision-making
Contributions:1 release, 30 pushes, 3 branches in 2 years 5 months
pytorchdecision-makingautonomousdecisionreinforcement-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.