Research Assistant

Generalist

8d•$150,000 - $200,000

About The Position

We are looking for a hands-on Research Assistant to help run real-world experiments at the intersection of robotics, machine learning, and physical-world evaluation. At Generalist, we are building foundation models for robots. These models improve through a tight feedback loop: collect data, train or fine-tune models, evaluate them in the real world, analyze results, and repeat. This role helps make that loop faster, more rigorous, and more reliable. You will work closely with ML researchers and robotics engineers to run robot experiments, design evaluation tasks, collect data, measure success rates, and document repeatable workflows. You do not need to be an experienced ML research scientist or robotics engineer, but you should be excited by careful experimentation, physical systems, statistical rigor, and hands-on iteration. A major part of this role is helping ensure our evaluations are trustworthy. We care deeply about experimental design, controls, sample sizes, variance, repeatability, and avoiding misleading conclusions from noisy real-world robot trials.

Requirements

Experience running experiments, lab studies, field studies, data collection workflows, or structured evaluations
Careful consideration of experimental design, confounding factors, controls, sample sizes, variance, and data support for conclusions
Diligence and detail-orientation, especially when tasks are repetitive but subtle differences matter
Enjoyment of hands-on work with physical systems, equipment, materials, or instruments
Comfort following protocols while also noticing when something is wrong or could be improved
Ability to coordinate many moving parts: robots, materials, tasks, data, model versions, metrics, and documentation
Clear communication skills, ability to summarize events, changes, and evidence-based suggestions
Curiosity about machine learning and robotics
Some exposure to programming, data analysis, robotics, hardware, electronics, mechanical assembly, or experimental tooling
Preference for fast iteration, careful measurement, statistical rigor, and empirical progress over abstract theory alone

Nice To Haves

This is not a pure ML Research Scientist role at the outset; you will not be expected to start by designing new model architectures or advancing core learning algorithms.
For someone with strong scientific judgment, technical curiosity, and excellent execution, it can grow into deeper research ownership over time, including proposing experiments, shaping evaluation methodology, and eventually contributing as a Research Scientist.

Responsibilities

Running structured experiments on robot platforms
Setting up physical tasks, materials, fixtures, and benchmarks for robot evaluations
Collecting high-quality robot data and tracking experimental conditions
Measuring real-world success rates across tasks, robots, and model variants
Designing evaluations with attention to controls, repeatability, statistical power, and sources of bias
Analyzing results to help distinguish real model improvements from noise
Synthesizing findings and communicating them clearly to ML researchers and engineers
Preparing robots, sensors, workspaces, and materials for rollouts and evaluations
Helping kick off training jobs, run evaluations, and organize results
Beta testing internal and third-party tools for teaching robots new skills
Troubleshooting physical setups, hardware issues, and procedural bottlenecks
Writing clear documentation and playbooks so others can reproduce workflows
Improving experimental reliability, data quality, and operational throughput over time