Vincent-Pierre Berges

• Created and released the Unity ML-Agents toolkit, an open-source toolkit that enables researchers to create simulation environments for Reinforcement Learning using the Unity Engine. Crossed 10,000 stars on Github in 2021.
• Continued development and support of the reinforcement learning algorithms of ML-Agents
• Examples of algorithms implemented include of Random Network Distillation, Hybrid Continuous and Discrete Actions, Attention Module, Centralized Critic, HyperNetworks for Goal Conditioning, Proximal Policy Optimization, Soft Actor Critic,
Generative Adversarial Imitation Learning, Intrinsic Curiosity Module and Behavioral Cloning
• Advised game studios integrating the ML-Agents toolkit
• Mentor to junior team members, including intern manager responsibilities
• Created a prototype of the ML-Agents toolkit using Unity’s Data Oriented Tools
Machine Learning Engineer May 2017 - April 2019 Unity Technologies, San Francisco, CA
• Wrote user friendly tutorial for game developers and researchers for the ML-Agents toolkit
• Featured in Forbes: 12 Amazing Deep Learning Breakthroughs of 2017