Senior RL Environments Engineer for Real-World Science

Lumos; San Francisco (CA); On-site; USD 160,000 - 220,000; Full time

Job summary

Lumos AI Research is seeking a Senior Software Engineer for RL Environments & Sandboxes to build high-fidelity environments where frontier AI agents learn to perform complex real-world tasks. You’ll design stateful, executable environments that interact with software, APIs, scientific tools, and lab systems, ensuring reproducibility and scalable evaluation.

You’ll work closely with our wet-lab team to connect AI agents to physical experiments, laboratory automation, and real biological outcomes,

Qualifications

• 5+ years of software engineering experience or equivalent demonstrated ability.
• Excellent Python skills and strong fundamentals in data structures, algorithms, systems design, and software architecture.
• Hands-on experience with Docker, Linux, containers, process isolation, or sandbox execution.
• Strong understanding of distributed systems, asynchronous execution, state management, and reliability.
• Experience designing complex APIs, backend systems, and integrations.
• Ability to independently architect, build, debug, and ship sophisticated technical systems.

Responsibilities

• Build RL environments: design and implement realistic, stateful environments across healthcare, biology, and scientific software.
• Develop secure sandboxes: create isolated, reproducible execution environments for agent-generated code and tool calls.
• Connect AI agents to wet lab workflows by integrating instruments, robots, protocols, and readouts.
• Build agent-facing tools: implement MCP servers, APIs, and interfaces for software interactions.
• Engineer state management: deterministic initialization, snapshot/restore, state transitions, resets, and replay.
• Develop tasks and verifiers: translate workflows into executable tasks with clear success criteria and ground-truth states.
• Scale environment execution: orchestration for high-throughput rollouts, including both simulations and physical experiments.
• Work with Harbor: validate environments using Harbor-compatible structures and evaluation workflows.
• Improve environment fidelity by identifying shortcuts and verification gaps.
• Own the full lifecycle from prototypes to production-grade infrastructure.

Skills

Python

Docker

Linux

Distributed systems

APIs

Software architecture

Algorithms

State management

Tools

Harbor

MCP

Gymnasium

Ray

Job description

Lumos AI Research is seeking a Senior Software Engineer for RL Environments & Sandboxes to build high-fidelity environments where frontier AI agents learn to perform complex real-world tasks. You’ll design stateful, executable environments that interact with software, APIs, scientific tools, and lab systems, ensuring reproducibility and scalable evaluation.

You’ll work closely with our wet-lab team to connect AI agents to physical experiments, laboratory automation, and real biological outcomes,