Adrien Ecoffet is a research scientist known for Go-Explore, an exploration algorithm that returned then explored from archived promising states, solving hard-exploration Atari games such as Montezuma's Revenge; a Nature paper (First return, then explore) presents the definitive version. He later joined OpenAI and coauthored work including Video PreTraining (VPT) and Weak-to-Strong Generalization.123
At a glance
- Current role
- Research scientist
- Company
- OpenAI
- Fields
- Reinforcement learning, Agents
- Sources cited
- 3
- Profile reviewed
Accomplishments
Go-Explore
Developed with collaborators at Uber AI Labs, Go-Explore archived promising states, returned to them, then explored—addressing sparse and deceptive rewards. The Nature article expands the approach with dynamic representations and robotics results.21
OpenAI
At OpenAI, Ecoffet appears on VPT Minecraft pretraining and as a coauthor of Weak-to-Strong Generalization.13
Timeline
Where they work
- OpenAI 102 people in the wiki
Related people
- Bowen BakerAlso at OpenAI
- Noam BrownAlso at OpenAI
- Kenneth StanleyAlso at OpenAI
- Christopher BernerAlso at OpenAI
- Ingmar KanitscheiderAlso at OpenAI
- Denis YaratsReinforcement learning · Agents
- David HaReinforcement learning · Agents
- Jakob FoersterReinforcement learning · Agents
- Manuel KroissReinforcement learning · Agents
- Niko GrupenAgents · Reinforcement learning
Suggest a correction or removal
Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.
References
3 sources · profile reviewed Oct 24, 2026 · data updated Oct 24, 2026