Principal Machine Learning Engineer (Memory & Context)

Role overview Lead the design and production use of memory and context systems for an AI assistant supporting extended home searches. The role determines how an assistant retains and updates useful information about each person, then assembles relevant context from conversation history, tools, and other sources so responses remain useful, timely, and efficient. It combines hands-on engineering with technical direction across teams.

Responsibilities - Set the architecture for user memory, including how information is written, retrieved, updated, and retired. - Design context assembly, prioritization, and compression for production agents. - Build and orchestrate agents from prompt and context design through deployment and ongoing operation. - Create offline and online evaluation methods for memory and context quality, and use results to guide product and technical decisions. - Track emerging agent approaches, including memory architectures and long-context methods, and apply them to product work. - Lead complex cross-team initiatives, communicate technical direction, and mentor engineers.

Requirements - Master's degree or higher in computer science, electrical engineering, or a related area, or equivalent experience. - At least five years deploying large-scale machine learning systems, including a recent year building and shipping LLM-based agents. - Practical experience developing agents end to end, including orchestration, prompt optimization, and context engineering. - Experience evaluating agent systems and using evaluation results to improve them. - Demonstrated technical leadership across teams and strong communication skills, including with senior leaders. - Strong Python skills and familiarity with agent frameworks such as LangGraph, Agents SDK, or AutoGen.

Nice to have - Experience with long-term, episodic, or user-profile memory; multi-session dialogue; or personalizing assistants from conversation history. - Publications, patents, or open-source work on memory, long-context models, or agents.

Benefits and work setup - Remote within the United States, with some location limitations. - Equity awards are available; compensation varies by location, experience, and performance.