Research Scientist
Generalist AI · San Mateo, CA · Somerville, MA
Train large pretrained robot models into production-ready systems through fine-tuning, reinforcement learning, human feedback and task specialization. Build evaluation frameworks and validate models on real robots, bridging ML and controls.
What you'll do
- Design fine-tuning and adaptation strategies for downstream robotic tasks and embodiments
- Develop methods to improve reliability, robustness and controllability
- Build evaluation frameworks measuring real-world robot performance
- Improve inference-time latency, stability and memory footprint with ML infrastructure
- Apply imitation learning, RL, distillation, synthetic data and curriculum learning
Nice to have
- Experience fine-tuning large models for downstream tasks (RLHF, IL, RL, distillation, domain adaptation)
- Experience with embodied AI, robotics or real-world ML systems
- Care about evaluation, benchmarking and failure analysis
- Comfortable debugging across the ML stack, from loss curves to robot behavior
- Enjoy rapid iteration with real-world feedback loops
Skills
Outlined skills are listed as required.
Full original description
About Generalist
At Generalist, we are on a mission to build general intelligence for the physical world and make it useful to everyone. We believe the industries and homes of the future will depend on humans and machines working together in new ways. Robots can help us build more and get more done.
We build embodied foundation models, starting with a focus on dexterity. This requires advancing the frontiers of data, models, and hardware, to enable robots to intelligently interact with the physical world.
The company embraces both large-scale AI and robotics as core to its DNA. Our team of researchers, roboticists, and company builders come from OpenAI, Boston Dynamics, Google DeepMind, and other frontier labs—with a track record of shipping AI breakthroughs. Before Generalist, we pioneered large embodied multimodal models and vision-language-action models (PaLM-E, RT-2, Gemini Robotics), launched and scaled ChatGPT and GPT-4 to hundreds of millions of users, engineered the foundations of autonomous driving, built next-generation robots (Atlas, Spot, Stretch) and pushed the limits of what they can do (from parkour to manipulation, and testing robustness).
We are an equal opportunity employer, and we do not discriminate on the basis of race, religion, color, national origin, sex, sexual orientation, age, veteran status, disability, genetic information, or other applicable legally protected characteristic.
About the Role
Pretraining gives us a general model. Post-training makes it useful, controllable, safe, and performant in the real world. You will train large pretrained robot models into production-ready systems via fine-tuning, reinforcement learning, steering, human feedback, task specialization, evaluation, and on-robot validation—at scale. Regardless of your initial background, you will grow into becoming a full-stack ML roboticist capable of quickly pinpoint issues on either side of ML or controls, and all the places in between. This is where research meets reality.
You’ll be responsible for:
Designing fine-tuning and adaptation strategies for downstream robotic tasks and embodiments
Developing methods for improving reliability, robustness, and controllability
Building evaluation frameworks that measure real-world robot performance, not just offline metrics
Improving inference-time performance (latency, stability, memory footprint) in collaboration with ML infrastructure
Leveraging techniques such as imitation learning, RL, distillation, synthetic data, and curriculum learning
Closing the loop between model outputs and physical-world outcomes
You might thrive in this role if you:
Have experience with fine-tuning large models for downstream tasks (RLHF, IL, RL, distillation, domain adaptation, etc.)
Have worked on embodied AI, robotics, or real-world ML systems
Care deeply about evaluation, benchmarking, and failure analysis
Are comfortable debugging across the ML stack — from loss curves to robot behavior
Enjoy rapid iteration with real-world feedback loops
Want to bridge the gap between foundation models and physical deployment