Platform Engineer, RL & Eval Data | 100% remote

Role overview Own and extend production platforms used to create evaluation tasks, reinforcement-learning environments, and human-reviewed data for advanced AI work. You’ll work across backend services, internal tools, and infrastructure, partnering with task authors and reviewers to turn operational needs into reliable software that can handle much larger workloads.

Responsibilities - Learn and maintain existing platforms: fix defects, reduce technical debt, and improve tests and documentation. - Design and deliver features across databases, backend services, APIs, and internal tools. - Scale systems for higher workloads with predictable performance and cost. - Build agent environments and supporting infrastructure, including isolated sandboxes, environment images, rollouts, and reward computation. - Maintain monitoring, alerting, incident response, access controls, client-data protections, and safe execution of untrusted code. - Create tools for task authors, reviewers, and project leads, including support for rapid sample production when needed.

Requirements - At least five years in software, data, or ML engineering, with recent experience in AI/ML data, model evaluation, annotation, or developer tooling. - A record of operating and maintaining production systems as well as building them. - Strong Python backend engineering, API and data-model design, and relational database skills such as schema changes and query tuning. - Enough TypeScript and React experience to build and maintain internal tools. - Experience with distributed systems for long-running jobs, orchestration and queues, containers, cloud infrastructure, and observability. - Security fundamentals, careful use of AI coding agents, initiative in ambiguous settings, and fluent English.

Nice to have - Hands-on work with LLM-agent RL environments, sandboxing, high-volume rollouts, evaluation harnesses, graders, or reward functions. - Experience with annotation workflows, contributor quality control, or scaling manual data processes. - Contributions to open-source RL or evaluation projects, or tools for distributed contributor teams.