About the role#
One Robot builds task-specific world models and an evaluation platform for robot manipulation policies. We replace trial-and-error development with rigorous validation that identifies policy failure modes and informs data collection. You will embed with our technical team to work on active research projects involving world models, policy training, and evaluation. This role is for those who want to work with physical robots rather than toy datasets.
What you'll do#
- Contribute to training runs by working with founding engineers on world model or evaluation model experiments.
- Run real-robot evaluations, which includes collecting demonstration data, testing policies on physical hardware, and documenting failure modes.
- Build tooling by writing Python and PyTorch code to improve our data engine, training pipelines, or evaluation infrastructure.
- Close the sim-to-real gap by running experiments that test how accurately simulation predicts real-robot behavior.
What you'll need#
- Current enrollment in a BS, MS, or PhD program in ML, robotics, computer vision, or a related field.
- Strong coding skills in Python and PyTorch.
- Hands-on experience training or fine-tuning a generative model, VLM, or policy through research or coursework.
- Ability to work in-person in San Francisco.
- Experience with real robot hardware or simulation environments such as Isaac or MuJoCo is a plus.
- Prior research in manipulation, 3D vision, or model evaluation is helpful.
Location & details#
- Location: San Francisco, California.
- Modality: On-site.
- Term: Rolling.
- Compensation: $6,000 - $8,000 per month.
- This is a full-time internship.
About One Robot (YC W26)
One Robot is a research services company based in San Francisco. Founded in 2025, the firm employs four people. It provides tools for robotics teams to automate data collection. The company uses world models to identify policy failures and create synthetic trajectories.
How to get in at One Robot (YC W26)
Securing an internship at One Robot requires speed, as early applicants are often reviewed before the candidate pool becomes unmanageable. Intern Insider sends an instant alert the moment a role matching your target is published anywhere, ensuring you apply among the first. Applying early provides a distinct advantage in a competitive field. You can also improve your response rates by reaching out to the recruiters behind the company roles directly. Intern Insider surfaces these contacts so you can inquire about the position or request a referral. This direct approach often yields better results than submitting an application into a general queue.
