The work
You will lead and mentor a pod of about 5 to 10 technical trainers who create and evaluate realistic terminal-based scientific and engineering tasks for AI training. You will own quality, delivery, documentation, work allocation, and technical review.
The work covers coding, systems, simulation, and electrical engineering tasks. You will check that tasks are accurate, reproducible, challenging, and measured with useful automated evaluations.
- Review problem statements, codebases, circuit netlists, hardware description languages, datasets, computing environments, reference solutions, and automated tests.
- Check algorithms, numerical accuracy, memory use, concurrency, circuit laws, power balance, signal processing, timing, and system stability.
- Test compilers, simulation tools, dependencies, build processes, and graders for reproducibility and meaningful results.
- Look for nondeterministic behavior and ways an evaluation could be bypassed.
- Give precise technical feedback, track revisions, resolve quality issues, and maintain technical best practices.
What it pays and takes
This is a part-time contractor role open worldwide. The input does not specify a pay rate. You should be prepared to commit at least 20 hours each week and work in English.
- Advanced technical experience through a Ph.D., postdoctoral work, or equivalent experience in computer science, electrical engineering, computer engineering, or a related field.
- Strong programming ability in C/C++, Rust, Python, Julia, Fortran, MATLAB or Octave, or hardware description languages.
- Advanced Linux and terminal skills.
- Hands-on expertise in parallel computing, systems programming, scientific software, circuit simulation, signal processing, power systems, digital hardware, or a closely related area.
- Strong knowledge of algorithms, computational systems, numerical methods, software testing, or electrical engineering principles.
- Experience reviewing complex technical work and mentoring or leading technical teams.
- Useful additional experience includes MPI, OpenMP, CUDA, LLVM or GCC, CMake, gdb, Valgrind, perf, Slurm, Docker, Git, CI/CD, pytest, simulation tools, automated grading, technical benchmarks, AI coding agents, open-source maintenance, or LLM-generated technical solutions.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.
About AI training work
AI training is the human work behind systems that learn from examples. Contributors create, review, and evaluate material such as code, technical tasks, and model responses, while experienced specialists help ensure the training data is accurate and useful.