Platform Engineer, RL & Eval Data (100% Remote)

About the role

The Human Data Division builds training and evaluation data for frontier AI labs, including RL environments, coding and STEM evaluation tasks and expert-annotated datasets. It has built its own evaluation, RL and annotation platforms and is looking for a platform engineer to join the platform team as a core member, helping keep the systems running, make them faster and more reliable, and build the features the team and its clients need. The role is 100% remote.

Key responsibilities

• Maintain and improve the platforms: learn the codebases in depth, fix bugs, reduce technical debt and improve test coverage and documentation.
• Build new features across the stack, from database and backend through APIs and internal tools.
• Scale the systems to much larger workloads with predictable performance and cost.
• Build RL environments and their infrastructure: sandboxes, environment images, rollouts and reward computation.
• Keep everything reliable and secure: monitoring, alerting, incident response, access control and protection of client data.
• Support the data teams by building the tooling used by task authors and reviewers, and producing fast-turnaround samples when needed.

Required profile

• 5+ years in software, data or ML engineering, with recent work in AI/ML data, model evaluation, annotation or developer tooling.
• A track record of running and maintaining production systems, not just building them.
• Fluent English.
• Nice to have: contributions to open-source evaluation or RL frameworks, and experience building tools for a large distributed contributor workforce.

Required skills

• Backend engineering in Python, API and data model design, relational databases (Postgres or similar).
• Frontend enough to build internal tools (TypeScript and React).
• Distributed system design, job orchestration, queues, containers and cloud (Docker, Kubernetes, at least one major cloud provider).
• Observability (logging, metrics, tracing, alerting) and security fundamentals.
• RL environments and sandbox infrastructure (Daytona, E2B, Modal, gVisor, Firecracker), environment images and rollout collection.
• Evaluation harnesses and benchmarks for LLMs, verifiers, graders and reward functions.
• Data annotation and human-data tooling, contributor and reviewer workflows.