6 hours ago
Responsibilities
- Define post-training recipes, including problem selection, measurement, data collection, machine-learning experiments, and production integration.
- Build and own post-training data pipelines using internal and public data.
- Create evaluations that measure whether models improve scientific protocols and assay development.
- Develop agentic systems with context management and custom tool calls for scientific experimental design.
- Collaborate with scientists, robotics engineers, and operations teams to integrate reasoning capabilities into live laboratory workflows.
- Help shape the engineering culture and technical direction of a new machine-learning team.
Requirements
- Practical experience building AI-driven workflows for real-world use.
- Strong problem-solving skills for debugging complex systems.
- Understanding of probability, statistics, and machine-learning fundamentals.
- Ability to own data pipelines, evaluation harnesses, reinforcement-learning environments, and agentic evaluations end to end.
- Proficiency in Python and familiarity with at least one deep-learning framework such as PyTorch or JAX.
- Experience with large language models, post-training, reinforcement learning, or agentic systems.
Benefits
- Opportunity to join an early-stage team building autonomous scientific laboratories and Physical AI systems.
- Collaboration with scientists, robotics engineers, and operations teams in real experimental environments.
- Mission-driven, collaborative engineering culture with team activities including skiing and go-karting.
Categories
AI ResearchML Engineering
About Medra
Medra builds Physical AI for laboratories: robotic systems and software that run experiments, analyze data, and iterate to accelerate R&D. It partners with biopharma companies to automate wet-lab workflows, offering an integrated hardware–software platform and services. Medra is privately held and headquartered in San Francisco.