Research Scientist, RL

Spectral Labs; San Francisco (CA); On-site; USD 300,000 - 400,000; Full time

Benefits offered by this job

Health insurance

Dental

401(k) match

Meals & commuting

Job summary

Spectral Labs is seeking a research scientist to lead reinforcement learning work for engineering design. You will build environments over CAD geometry, design rewards grounded in geometric validity, and supervise simulation and post-training at scale.

The role requires 7–10 years of research experience in RL, ML, or robotics, with strong fundamentals and the ability to write training code yourself. This is a fully in-person position in our SF office, offering total compensation and equity in

Qualifications

• 7–10 years of RL/ML/robotics research experience.
• Strong fundamentals and ability to write training code.

Responsibilities

• Lead reinforcement learning research for engineering design.
• Build environments over CAD geometry and design rewards grounded in geometric validity.
• Scale post-training of models and simulations across compute resources.

Skills

RL research experience

Machine learning

Robotics

Training code implementation

Job description

Spectral Labs is building foundation models for engineering physical systems. We're looking for a research scientist to lead our work on reinforcement learning for engineering design: building environments over CAD geometry, designing rewards grounded in geometric validity, simulation, and manufacturability, and post-training our models at scale. You bring 7–10 years of research experience in RL, machine learning, or robotics, strong fundamentals, and the implementation skills to write the training code yourself. We're excited about people who can make an impact, whatever their background.

This is a fully in-person role at our office in SF. Compensation and equity total $300–400k annually. We sponsor visas end-to-end, and benefits include:

• health insurance (with a 100%-premium option)
• dental
• a 401(k) match
• company-expensed meals and commuting.