Research Scientist Intern, Video RecSys (PhD)

Meta•Menlo Park, CA

6d•$7,650 - $12,667

About The Position

Meta is seeking a PhD-level AI Research intern to join Facebook’s Video Recommendations team. The team recommends Reels, long-form videos, and Live streams using large-scale unified video recommendation models. The intern will work on Recursive Self-Improvement (RSI), building AI agents that autonomously propose, run, evaluate, and learn from improvements to recommendation models. The work is intended to connect machine-learning experimentation with model quality and will be mentored by Research Scientists and Software Engineers. Production changes and online experiments require review and approval by owners. The intern will develop an agent-driven RSI prototype for a selected video recommendation task. The project may span training-data improvement, environment and experiment setup, automated offline model evaluation, and collection of model-quality signals such as normalized entropy (NE) and recall. Experiment outcomes will be used to improve the agent’s intervention strategy and evaluation criteria. The goal is to deliver a working end-to-end prototype with evidence against a fixed baseline. Internships last 12 to 24 weeks, with start dates throughout the year.

Requirements

Currently has or is in the process of obtaining a Ph.D. in Computer Science, Artificial Intelligence, Machine Learning, or a relevant technical field.
Must obtain work authorization in the country of employment by the time of hire and maintain ongoing work authorization during employment.
Experience with Python, C++, C, Java, or another related programming language.
Experience with deep-learning frameworks such as PyTorch or TensorFlow.
Familiarity with machine learning and common deep-learning architectures.

Nice To Haves

Intent to return to the degree program after completing the internship or co-op.
Research achievements demonstrated through grants, fellowships, patents, or publications at leading AI conferences such as NeurIPS, ICML, ICLR, KDD, RecSys, WWW, ACL, or EMNLP.
Research or project experience in one or more of the following: LLM-based agents and agentic workflows, recursive self-improvement, automated machine learning (AutoML), data-centric AI (data selection, curation, and reweighting), reinforcement learning, or large-scale recommendation systems.
Experience building reproducible ML experimentation and evaluation pipelines on large-scale distributed training systems.
Software-development experience through technical internships, work experience, coding competitions, or widely used contributions to open-source machine-learning repositories.
Experience working and communicating cross-functionally in a fast-paced team environment, with the ability to identify research opportunities in real-world applications, select appropriate ML methods, and proactively drive iterations based on analysis of results.

Responsibilities

Design and build an agent-driven Recursive Self-Improvement system that continuously proposes, evaluates, and refines improvements to a large-scale video recommendation model.
Profile training-data quality and signal coverage, and implement bounded data interventions such as filtering, sampling, and reweighting.
Automate environment and experiment setup, including reproducible training configurations, offline model-evaluation pipelines, and evaluation benchmarks.
Build feedback loops to collect and attribute model-quality signals, such as normalized entropy and recall gains or losses, and use them to improve the agent’s intervention strategy and evaluation criteria.
Measure downstream model quality alongside data-processing and training costs; maintain versioned experiment records and automated checks.
Collaborate with researchers and cross-functional partners, communicating research plans, progress, and results.
Publish research results and contribute to research applicable to Meta product development.