Research Scientist Intern, AI Core Machine Learning (PhD)

Job description

Meta was built to help people connect and share, and over the last decade, our tools have played a critical part in changing how people around the world communicate with one another. With over a billion people using the service and more than fifty offices around the globe, a career at Meta offers countless ways to make an impact in a fast growing organization. Meta is seeking a Research Intern to join Monetization Ranking & AI Foundations to answer the question: can an AI research agent get better at research by learning from its own experiments? This project explores recursive self-improvement for autoresearch: AI agents that form hypotheses, run experiments, analyze results, and then use that experience to improve how they conduct future research. Our team develops an agentic ML-research system that proposes, trains, and evaluates model changes at production scale, continually accumulating real research trajectories and outcomes from which the system can learn. The intern will help shape the research direction around open questions such as: Which parts of an agent should learn from experience—the LLM's weights through reinforcement-learning post-training, its harness of prompts, tools, memory, and orchestration, or both? How can we build reliable reward signals when real experiments are slow and expensive, while preventing reward gaming? Does self-improvement compound across rounds, or does it eventually plateau or collapse? Our interns closely collaborate with researchers and engineers within the organization and across the company (e.g., FAIR), push the boundaries of state-of-the-art research in AI and ML, and integrate innovations at unprecedented scale. The intern will have room to define their own research question within this space, with the goal of producing a full paper for ICLR, ICML, or NeurIPS. Our internships are twelve (12) to sixteen (16), or twenty-four (24) weeks long and we have various start dates throughout the year.