RL Environments Engineer

1 month ago

Mountain View, CA, USASenior

H1B sponsor

Base Salary

$250k - $300k/yr

Responsibilities

Build environment-generation pipelines for templating, automated grading, verification, and QA.
Create high-fidelity coding environments based on real codebases, dependencies, tooling, and technical debt.
Scale agentic coding-task generation to hundreds or thousands of validated tasks through automation.
Build internal tools and infrastructure that remove production bottlenecks and increase team throughput.
Own prompts, environments, graders, frontier-model execution, failure analysis, and task iteration.
Develop verification standards and defenses against reward hacking and grader loopholes.
Use frontier coding agents to build and validate environments, assess their output, and identify subtle failures.

Demonstrated experience shipping agentic coding tasks or RL environments at meaningful volume, including the number of tasks and production cost.
Experience scaling task or environment output through automation rather than additional manual labor.
Strong production software engineering fundamentals and fluency in several programming languages.
Experience with large codebases, build systems, testing, deployment, on-call, and root-cause analysis.
An adversarial approach to identifying how models could exploit graders and fixing those weaknesses.
Clear understanding of the capabilities and limitations of frontier coding agents.
Ability to build, debug, and ship with limited supervision.
Experience with RL training systems, post-training, verifiers, or tool-use harnesses is a plus.
Background in developer tooling, CI/CD, sandboxes, or code-execution infrastructure is a plus.
Experience with automated test generation, fuzzing harnesses, benchmark suites, public agentic benchmarks such as Terminal-Bench, or relied-upon open-source projects is a plus.

1-10 employees

RL for Agents