Bespoke Labs Seeks Remote Contract Researcher for Long-Horizon AI Agent Benchmarks

Bespoke Labs, a frontier AI company based in Mountain View, is hiring a contract researcher on a remote basis to design and evaluate reinforcement learning environments for long-horizon agentic tasks. The role focuses on building RL benchmarks grounded in real-world workflows such as coding, tool use, and enterprise processes — tasks requiring multi-step reasoning over hours, days, or weeks. Applicants must demonstrate prior hands-on experience with long-horizon agent or RL systems, such as contributions to benchmarks like SWE-bench or Gymnasium; applications lacking this evidence will not be considered. Strong Python skills and familiarity with RL training and evaluation frameworks are also required. Interested candidates should apply by submitting a brief note along with verifiable links to relevant work, including GitHub repositories, research papers, or production systems.