• Build and maintain scrapers for problems and submissions (AC, WA, TLE, MLE, RE).
• Generate test suites using testlib across full constraint ranges.
• Validate tests by ensuring accepted solutions pass, incorrect solutions fail, and that test suites are robust enough to verify the intended algorithmic approach rather than just output correctness.
• Build and maintain the grading harness for compilation, execution, resource enforcement, and verdict generation.
• Run model benchmarks and package evaluation results.

‍

• Ability to work 2:00 p.m. - 10:00 p.m. (CEST)
• 4+ years as a Python Engineer (data pipelines, orchestration, APIs, automation, and tooling)
• Experience running untrusted code with resource limits, process isolation, timeouts, and execution monitoring
• Ability to read, compile, and debug C++ code
• Skills in building multi-stage pipelines (ingest → normalize → generate → validate → package), handling large datasets, schema validation, and deduplication
• Ability to work with API Integrations (LLM APIs, rate limits, retries, concurrency, and batch processing)
• Verification Mindset - strong focus on correctness, validation, testing, and benchmarking.
• Docker and containerized execution environments

Nice to have:

• BS/MS in Computer Science, Mathematics, Statistics, or a related quantitative field
• Some experience with AWS (EC2, S3, batch processing)
• RL training data or evaluation infrastructure knowledge
• Experience in HuggingFace Datasets
• Skills in the online judge systems (Codeforces, AtCoder, LOJ)