Job Description

Meta's mission is to give people the power to build community and bring the world closer together, and the ATA organization focuses on delivering this vision through AI.

Meta's Memory team within the ATA organization is seeking a Research Scientist Intern with a focus on LLM memory. As a Research Scientist Intern, your primary objective is to develop cutting-edge memory algorithms and production-ready memory systems for large language models (LLMs). We are seeking candidates to work on fundamental improvements to LLM memory, with a strong focus on scalability for deployment at Meta scale. Our internships are twelve (12) to twenty-four (24) weeks long and we offer various start dates throughout the year.

Responsibilities

Perform research on LLM memory that leads to deployment at meta scale and potentially publications in top-tier NLP/ML conferences

• Develop new methods for long-term memory in LLMs and LLM agents, such as memory formation, retrieval, consolidation, and forgetting
• Build production-ready memory systems that scale to large numbers of users and large volumes of data, with low latency and low cost
• Design benchmarks and evaluation methods that measure memory quality, such as recall accuracy, relevance, freshness, and personalization
• Design benchmarks and evaluation methods that measure system performance, such as latency, throughput, and cost
• Collaborate with cross-disciplinary research and engineering teams to bring research into products

Qualifications

Currently is in the process of obtaining a PhD in the field of Artificial Intelligence or related field

• Must obtain work authorization in the country of employment at the time of hire, and maintain ongoing work authorization during employment
• Hands-on experience with LLMs, such as pretraining, fine-tuning, post-training, retrieval-augmented generation (RAG), benchmarks or agentic workflows
• Solid understanding of transformer architectures and modern deep learning techniques
• 3+ years experience with deep learning software libraries such as PyTorch/JAX
• Experience with scripting languages such as Python and shell scripts Intent to return to degree-program after the completion of the internship
• Research experience in one or more of these areas: long-term memory for LLMs or agents, RAG, knowledge editing, long-context modeling, or personalization
• Experience with building and deploying machine learning systems in production
• Experience with large-scale distributed systems for LLM training, inference, or retrieval
• Experience with building evaluation frameworks or benchmarks for LLMs
• Proven track record of publications at ML/NLP conferences/journals such as NeurIPS, ICLR, ICML, ACL, EMNLP, NAACL, COLM, AAAI
• Interpersonal experience: cross-group and cross-culture collaboration