RL Environment Engineer
Scraped from: Afriworket
Our AI company looking for a talented and motivated Reinforcement Learning Environment Engineer to join our team. In this role, you will design, build, and maintain simulation environments that serve as the training ground for our RL agents — bridging the gap between research ideas and scalable, production-ready solutions. What you will build Sandboxed, reproducible environments that wrap real software workflows (browser, terminal, IDE, internal tools) so an agent can act inside them end to end. Reward and verification functions that score agent behavior against ground truth rather than heuristics. Eval harnesses that prove an environment actually separates strong policies from weak ones. Tooling that converts a captured expert trajectory into a trainable task with the least possible loss of fidelity. What we're looking for Strong Python and the ability to build and debug non-trivial systems on your own. Working knowledge of RL fundamentals: policies, rewards, rollouts, on-policy vs off-policy. You do not need to be a researcher, but you should understand why a reward function is hard to get right. Experience with containerization and sandboxed execution (Docker, gVisor, Firecracker, or similar). Familiarity with LLM agents and tool use: function calling, agent loops, browser and terminal automation. A track record of building verifiers, graders, or test harnesses where correctness genuinely matters. Comfort with ambiguity and messy, real-world workflows. For those interested, please use this link to apply. https://forms.gle/XwmxerisorXd9oHU8