Research Scientist, Program Synthesis & Neuro-symbolic Methods

What you'll do

• Develop theories and algorithms for synthesizing control programs, learning interpretable models, and combining symbolic reasoning with neural learning.
• Integrate research across world modeling, planning, reasoning, and control; validate algorithms on physical robots.
• Build and maintain open-source software and collaborate on applications of MARA technology.

What they ask for

• Demonstrated high-quality scientific research, such as publications, technical reports, or impactful software projects.
• Deep expertise in program synthesis and neuro-symbolic methods, including experience combining neural networks with symbolic reasoning or program generation.
• Strong foundations in probability theory, optimization, and linear algebra, with the ability to implement complex algorithms from first principles.
• Ability to work across theoretical development and experimental validation.

Nice to have

• PhD or equivalent experience in a technical area such as robotics, machine learning, computer vision, control theory, cognitive science, or physics.
• Experience at leading robotics or AI labs; deployment on physical robots; major robotics or ML open-source contributions; or experience spanning theory and systems engineering.

Pay

$120K – $180K (currency not stated; salary range)

Tools

Domain-specific languages, program induction, verifiable control, neural networks

AI focus

Research and develop program synthesis and neuro-symbolic methods for embodied intelligence, including interpretable and verifiable robot control.