Research Engineer (Robotics / Foundation Models)
Research Engineer responsible for scaling ML experiments and infrastructure for robotics and foundation model development. Partner with researchers to implement architectures, manage data pipelines, maintain distributed training stacks, and evaluate models in simulation and on physical robots. Requires 2+ years ML engineering experience with PyTorch or JAX, distributed training, and debugging across GPU clusters.
Key Highlights
Key Responsibilities
Technical Skills Required
Benefits & Perks
Nice to Have
Job Description
We are running this search on behalf of our client.
Research Engineer (Robotics / Foundation Models)
Full-time · On-site, San Francisco · $250K–$400K + equity
How fast the company progresses comes down to how many experiments the team can run each week, and Research Engineers are the people who set that pace. You'll partner with researchers across pretraining and RL post-training, build the data and infrastructure that let those runs scale, and stay involved all the way from the GPU cluster out to robots operating on real factory floors.
What the work looks like:
- Turning model architectures and training recipes into something that runs well at scale, hand in hand with the research team
- Standing up and maintaining the data side: gathering, cleaning, filtering, and augmenting data spanning vision, proprioception, action, and language
- Keeping the training stack healthy, including distributed runs, checkpointing, profiling, and debugging across a large GPU fleet
- Writing evaluation that surfaces real regressions, both in simulation and on physical robots
- Owning the full cycle from training to a deployed robot, then feeding field data back in
Searching for Development & Programming roles that provide visa sponsorship? Connect with international employers through Development & Programming Jobs with Visa Sponsorship opportunities actively seeking talented professionals.
Who tends to do well here:
- Engineers first, with strong ML instincts
- People who have built ML systems end to end rather than only operating existing ones
- Comfortable in PyTorch or JAX, with real distributed training experience
- Able to debug up and down the stack
- Productive when the problem is still ambiguous
Nice to have:
Explore our comprehensive directory of visa sponsorship jobs from employers worldwide who are ready to sponsor talented international professionals.
- Time spent on multi-node, multi-GPU training setups
- Work on large multimodal models
- Papers at venues like NeurIPS, ICML, ICLR, CoRL, RSS, or ICRA
- Shipping models onto real hardware and tuning them for edge latency and compute
The basics:
- Roughly 2+ years in ML engineering, ideally touching large-scale training
- $250K–$400K base plus meaningful equity
- Visa sponsorship possible for most cases, though you'd need to start soon
- Fully on-site, five days a week in San Francisco
- Stack: PyTorch, JAX, Python, CUDA, distributed training, H100s, Linux
Similar Jobs
Explore other opportunities that match your interests