Embodied AI Scientist: Post-Training to Real-World Robotics

Generalist; Medford Hillside (MA); On-site; USD 180,000 - 240,000; Full time

Job summary

Generalist is building general intelligence for the physical world, focusing on dexterity and production-ready robotic systems. You will train large pretrained robot models into scalable, production-ready implementations through fine-tuning, RL, and human feedback, bridging research and real-world deployment.

Joining requires expertise in fine-tuning, embodied robotics, and robust evaluation. You’ll work across ML and controls, improving reliability and performance in real-world tasks and

Qualifications

• Experience fine-tuning large models for downstream tasks (RLHF, IL, RL, distillation, domain adaptation).
• Background in embodied AI, robotics, or real-world ML systems.
• Strong focus on evaluation, benchmarking, and failure analysis.
• Ability to debug across the ML stack from training curves to robot behavior.
• Comfort with rapid iteration using real-world feedback loops.

Responsibilities

• Design fine-tuning and adaptation strategies for downstream robotic tasks and embodiments.
• Develop methods to improve reliability, robustness, and controllability.
• Build evaluation frameworks that measure real-world robot performance, beyond offline metrics.
• Improve inference-time performance (latency, stability, memory footprint) with ML infrastructure collaboration.
• Leverage imitation learning, RL, distillation, synthetic data, and curriculum learning.
• Close the loop between model outputs and physical-world outcomes.

Skills

Fine-tuning large models

Embodied AI experience

Robotics/ML systems

Evaluation & benchmarking

Debugging ML stacks

Rapid iteration & productionization

Job description

Generalist is building general intelligence for the physical world, focusing on dexterity and production-ready robotic systems. You will train large pretrained robot models into scalable, production-ready implementations through fine-tuning, RL, and human feedback, bridging research and real-world deployment.

Joining requires expertise in fine-tuning, embodied robotics, and robust evaluation. You’ll work across ML and controls, improving reliability and performance in real-world tasks and