Research Scientist Intern - PhD — Meta Superintelligence Labs

Meta is seeking Research Interns to join our AI Search research team. The team advances artificial intelligence by developing methods for AI agents to learn from their own experience, improve search and reasoning workflows, retain useful knowledge, and transfer those gains to unfamiliar tasks. The intern will work in one of three directions: - Learning to improve research workflows: Agents learn better search, tool-use, and verification strategies from prior attempts by improving their harness, including the workflow, memory, tools, and feedback loop around the model. Topics include meta-learning across tasks, automated agent design, and learning from natural-language feedback. - Persistent memory and continual learning: Agents retain useful lessons and update knowledge across tasks, whether in context, in compressed memory, or in weights, without forgetting what they already know. Topics include long-context modeling, context compression, continual learning, and test-time training. - Learning from imperfect feedback: Turning noisy judgments and failed attempts into useful learning signals and improving the strategy an agent uses to learn, including how it selects and evaluates its next improvement. Topics include reward learning, credit assignment, and learning to learn.Across all three directions, the key test is whether improvements transfer to new tasks. The goal is to develop novel methods and experiments toward a first-author submission at a leading ML/NLP venue. Recent internships on this team have produced work such as REFRAG, a new architecture for retrieval-augmented generation. Internships are twelve to twenty-four weeks long, with various start dates throughout the year. ## Responsibilities - Develop novel state-of-the-art methods for self-improving, agentic AI systems, leveraging deep learning and reinforcement learning techniques. - Design and run experiments that test whether agent improvements generalize and transfer to unseen tasks. - Perform research to advance the science and technology of intelligent machines. - Help analyze and improve the reliability and robustness of the resulting methods. - Collaborate with researchers and cross-functional partners, including communicating research plans, progress, and results. - Disseminate and publish research results, and contribute to research that can be applied to Meta product development. ## Minimum Qualifications - Currently has or is in the process of obtaining a Ph.D. degree in Computer Science, Artificial Intelligence, Machine Learning, Natural Language Processing, or a relevant technical field. - Must obtain work authorization in the country of employment at the time of hire and maintain ongoing work authorization during employment. - Research experience in machine learning, deep learning, and/or natural language processing. - Experience with Python. - Experience with deep learning frameworks such as PyTorch or TensorFlow. ## Preferred Qualifications - Intent to return to the degree program after completion of the internship or co-op. - Significant research results demonstrated through grants, fellowships, patents, or first-authored publications at leading workshops or conferences such as NeurIPS, ICLR, ICML, ACL, EMNLP, or COLM. - Research experience in large language models, foundation models, or post-training. - Experience with agentic systems, including agent workflows, tool use, agentic search, or retrieval-augmented generation. - Experience with reinforcement learning, reward modeling, or credit assignment. - Experience with continual or lifelong learning, meta-learning, or test-time training. - Experience with long-context modeling, memory, context compression, or efficient inference. - Experience with LLM evaluation, setting up machine learning experiments, and analyzing results. - Experience with Hugging Face Transformers and training or fine-tuning transformer models. - Experience with distributed training. - Experience working and communicating cross-functionally in a team environment. - Demonstrated software engineering experience through an internship, work experience, coding competitions, or widely used contributions to open-source repositories.