Research Scientist Intern — Multi-Agent Memory & Reasoning

Meta; Bellevue, Seattle, New York (WA, WA, NY); On-site; USD 34,000 - 55,000; Part time

Job summary

Meta is seeking Research Interns to join our FAIR Communication & Language teams in Bellevue, WA for 12–24 weeks. You will work on multi-agent systems, memory and language models, and fundamental AI breakthroughs.

We welcome PhD candidates with experience in Python, PyTorch, RL, and large language models, and a track record of research results. You will prototype ideas at scale and contribute to core algorithms.

Qualifications

• Pursuing a PhD in CS/ML/AI/NLP with strong research
• Experience in large language models, ML, RL
• Experience with Python, C++, or related languages
• Authorization to work in the United States
• Proven research results (grants, publications) at leading AI conferences
• Research experience in agent memory, context compression, multi-agent systems
• Experience building language-model agents with tools, memory, or collaboration
• Experience with PyTorch and prototyping research code
• Contributions to open source libraries are welcome
• Ability to design controlled experiments and analyze trajectories

Responsibilities

• Investigate how AI agents retain knowledge across extended interactions
• Develop compact memory representations and retrieval strategies
• Improve inter-agent communication and coordination mechanics
• Explore automatic optimization of agent harnesses and workflows
• Evaluate ideas at scale and communicate findings

Skills

Multi-agent systems

Agent memory

Long-context modeling

Self-improving agents

Reinforcement learning

Python

C++

Research experience

Education

PhD in CS/ML/AI/NLP

Tools

PyTorch

Open source contributions

Job description

Meta is seeking Research Interns to join our FAIR Communication & Language teams in Bellevue, WA for 12–24 weeks. You will work on multi-agent systems, memory and language models, and fundamental AI breakthroughs.

We welcome PhD candidates with experience in Python, PyTorch, RL, and large language models, and a track record of research results. You will prototype ideas at scale and contribute to core algorithms.