Visiting Scholar: Post-Training & Agentic AI Research

Meta; Menlo Park, Northern (CA, KY); Hybrid; USD 230,000 - 301,000; Full time

Job summary

Meta's Discovery AI Lab in Menlo Park, CA invites applications for a Visiting Scholar to join a 12-month, full-time research engagement focused on post-training and generation. You will work with a world-class team, leveraging Meta-scale infrastructure to advance LLMs in recommendation system contexts.

Responsibilities include leading post-training algorithm research (RLHF, preference optimization), designing self-improving agent frameworks, and mentoring researchers, with a path to

Qualifications

• PhD in CS/ML/NLP or related field required.
• Active university or research appointment required.
• Strong publication record in top venues.

Responsibilities

• Lead research on post-training algorithms for generative recommendation systems.
• Design and develop self-improving agent frameworks using multi-agent collaboration and continual learning loops.
• Advance efficient inference techniques (quantization, compression, distillation) for large-scale models.
• Collaborate with researchers and engineers to translate research into production-ready systems at Meta scale.
• Mentor researchers to uplift internal capabilities in post-training and agentic AI.

Skills

Research collaboration

Mentoring

Publication track record

Team communication

Education

PhD in Computer Science/ML/NLP

Active faculty appointment

Bachelor's degree in CS/Engineering

Tools

Post-training methods

Large language models

Agentic AI systems

Job description

Meta's Discovery AI Lab in Menlo Park, CA invites applications for a Visiting Scholar to join a 12-month, full-time research engagement focused on post-training and generation. You will work with a world-class team, leveraging Meta-scale infrastructure to advance LLMs in recommendation system contexts.

Responsibilities include leading post-training algorithm research (RLHF, preference optimization), designing self-improving agent frameworks, and mentoring researchers, with a path to