The work
You will design reproducible reinforcement learning environments around complex software engineering workflows. Typical tasks resemble real DevOps, CI/CD and debugging work using tools such as git, docker, gdb, asan and ffmpeg. Each environment needs a clear objective, controlled setup, reliable evaluation criteria and a golden reference solution that demonstrates the intended path.
Technical scope
Required language proficiency is in C++, Python, JAVA, GoLang, Typescript, or Rust. The work draws on algorithms, data structures, performance tuning, feature development, refactoring and difficult debugging. Experience with distributed codebases, rigorous code review or AI and machine learning systems is useful, but AI experience is not required.
Screening process
The process begins with application questions, followed by an AI interview of approximately 30 minutes. A technical assessment may follow before hiring manager review. Expect the screen to press for verifiable examples of complex failures you diagnosed, performance or scalability improvements you measured, and how you would make a model-evaluation environment deterministic, difficult and objectively gradable.
Logistics and compensation
The contract is remote and open globally, with a typical commitment of about 15 hours per week. Work is output-based, accepted tasks must meet project specifications, and minimum weekly submission requirements apply, although the listing does not state the number. The listing does not promise a fully asynchronous schedule, so confirm review and handoff windows during onboarding; selected experts are expected to begin their first tasks within 24 to 48 hours after onboarding. Pay band: $80–120/hr
This band is observed from the listing and is not guaranteed.