ABOUT BESPOKE LABS

Bespoke Labs is an applied AI research lab pioneering data curation and RL environment curation for the modern agentic world. We recently curated one of the best open reasoning datasets, used by multiple frontier labs. We also trained SOTA specialized models such as Bespoke-MiniChart-7B and Bespoke-MiniCheck with meticulous data curation. See also our recent work on using RL to train mutli-turn tool calling agents.

ABOUT THE ROLE

As a member of our technical staff, you will work on all aspects related to data curation and RL environment curation, starting from manually curating environments, designing recipes, and working with skilled contractors. Ideal candidates are problem solvers who can understand the problem in a scientific way and can solve the problem practically.

WHAT YOU WILL DO

• Build our curation platforms for building/collecting/curating RL environments and data curation.
• Do research on cutting edge curation strategies, especially for RL environments.
• Come up with data and environment recipes, and work with contractors to create RL environments.
• Verify whether environments are high quality, by checking for reward hacking, and training small scale agents.
• Do data analysis to uncover insights about the environments.