About this role
Boston Dynamics’ Atlas Applications team seeks a Staff Machine Learning Engineer to build the agentic planning and reasoning layer, called System 2, above humanoid robot control. The goal is to enable Atlas to carry out long-horizon tasks in dynamic environments using task prompts, planning, vision-language model inference, memory and tool use.
Responsibilities
- Design and implement System 2 architectures for planning, reasoning, memory and tool use in humanoid robot control.
- Build logging, observability and evaluation methods to improve performance and assess reasoning quality, safety, reliability and generalization.
- Work with behavior, controls, perception and product teams to deliver end-to-end capabilities. Engage customers and internal stakeholders to understand real-world use cases.
- Apply advances in agentic research to production systems, and contribute through code reviews, testing, documentation and system design.
Requirements
- 5+ years of professional software engineering experience, including significant work on LLM-driven or agentic systems.
- Hands-on experience building or deploying agentic architectures, such as coding agents, tool-using LLM systems or autonomous task agents, and a record of improving performance through evaluation and benchmarking.
- Strong foundations in data structures, algorithms, distributed systems and software architecture; ability to ship reliable, maintainable, well-tested software and reason about safety, failure modes and robustness in autonomous systems.
Preferred
Experience with robotics, vision models and vision-language-action models; proficiency in TypeScript, Python and/or C++ across large codebases; experience applying coding agents; and research, open-source or production contributions that advance the state of the art.
The full-time role is listed at Waltham Office (POST). Annual base pay is $155,000–$235,000, depending on individualized factors. Benefits include medical, dental and vision coverage, a 401(k), paid time off and an annual bonus structure.