Lumen Labs | Simulation/RL Integration Engineer
Location: REMOTE (US/Canada), transitioning to ONSITE in San Francisco from Fall 2026 Compensation: $130k–200k + early equity Visa: Sponsorship available
About Lumen Labs
Lumen (lumenresearch.ai) is building the cognitive layer for physical AI. We believe that the current reliance on thousands of hours of human teleoperation for robotics has a performance ceiling.
- Read our white paper: lumenresearch.co/paper
- Team: We are a team of two ex-founders; you would be the third hire.
- Status: Pre-seed, with runway well into 2027 and a seed round planned for later this year.
The Role
We test many model architectures against many simulations, and integrating them is our primary technical challenge. You will own our evaluation harness and be responsible for:
- Building per-simulation adapters to a shared interface contract.
- Bridging single-environment vs. GPU-vectorized and sync vs. async paradigms.
- Using Gymnasium and Shimmy as the core plumbing.
- Contributing upstream improvements to open-source projects.
Requirements
- Strong Python proficiency.
- Deep familiarity with Gymnasium, specifically the vector API.
- Hands-on experience with Isaac, MuJoCo, or Gazebo.
- A proven track record of successfully getting models running in simulations where they were not originally built to run.
How to Apply
Email hi@lumenresearch.co with information about yourself and your projects.
Pro-tip: Link a project where you successfully got a model working in a simulation it wasn't built for to jump to the top of the pile.
Full Posting: Notion Link