Research Engineer / Scientist, Memory
About this role
Role Overview
The Research Engineer / Scientist, Memory position at Letta is an in-person role located at the San Francisco office, requiring five days a week on-site. This opportunity suits an experienced AI professional driven by high-stakes agent research and eager to design advanced long-term memory frameworks for large language models.
Letta is building self-improving artificial intelligence designed to help agents learn continually from daily experiences. Founded by the creators of MemGPT from UC Berkeley’s Sky Computing Lab and backed by prominent AI pioneers, the company focuses on creating adaptive systems that power real-world production environments for top companies. Joining this tight-knit, flat organization places team members on the front lines of solving major artificial intelligence challenges without bureaucratic management layers.
Key Responsibilities
Requirements
Skills and Tools
About the Company
Letta originated from UC Berkeley’s Sky Computing Lab, the historic birthplace of Ray and Spark. The organization is backed by industry leaders including Clem Delangue and Jeff Dean. Letta agents currently run in production environments for enterprise clients such as Bilt Rewards and 11x. The company champions open-source values and opposes closed frontier models controlled by a select few, offering a high-velocity startup environment where individual contributors maximize their personal impact.
How to Apply
Interested candidates should visit the official company careers page to submit their application for the Research Engineer / Scientist, Memory role.
Skills and tech stack
Questions candidates ask
Where is the Research Engineer, Memory role located?
This position is strictly in-person, requiring attendance five days a week at the downtown San Francisco office with no remote or hybrid options.
What is the primary focus of the Research Engineer, Memory position?
The role centers on designing the long-term memory architecture for large language models, including context management, memory sharing across multiple agents, and semantic knowledge storage.
What background is required for applicants?
Applicants need deep expertise in retrieval and large language models, a history of impactful research through publications or open-source work, and the ability to balance empirical rigor with execution speed.
What does the interview process look like?
The hiring pipeline includes a 30-minute initial screen, a technical screen lasting one to one-and-a-half hours, and a paid two-day in-person work trial onsite in San Francisco.
How is the team structured at Letta?
Letta operates as a flat organization where every researcher and engineer functions as an individual contributor without people-management duties or traditional corporate hierarchies.