Osama Hanna

Osama Hanna

Research Scientist at Meta Superintelligence Lab

Ph.D. UCLA '24

About Me

I am a Research Scientist at Meta Superintelligence Lab (MSL) in Menlo Park. My current work focuses on Post-training and Reinforcement Learning from Human Feedback (RLHF) for large language models, including leading efforts for post-training the audio Llama model family.

I completed my Ph.D. in Electrical and Computer Engineering at UCLA (2024), advised by Prof. Christina Fragouli and Prof. Lin F. Yang. My doctoral research on online learning and information theory was recognized with the UCLA Distinguished Ph.D. Dissertation Research Award (2024-2025).

My background is in mathematics and theoretical engineering. During my time at UCLA, I dove deep into graduate-level optimization, probability theory, and mathematical logic (including Gödel’s theorems and axiomatic set theory); early in my journey, I ranked 1st in the Egyptian Math Olympiad. Today, I enjoy connecting those theoretical concepts to practical, large-scale AI alignment and efficiency problems.

Research Interests

GenAI & LLMs

Post-training, RLHF, Alignment, and Efficient Inference.

Reinforcement Learning

MDPs, Bandits, Regret Minimization, and Sample Efficiency.

Information Theory

Fundamental Limits of Learning and Lower Bounds, Compression.

Optimization

Theory and Applications.

News

Selected Publications

Loading publications...

Invited Talks

Service & Mentoring

Mentoring

I have had the privilege of mentoring Ph.D. students (Merve Karakas, Xinlin Li, Kaan Buyukkalayci) and undergraduates at UCLA. At Meta, I manage research interns (e.g., Yancheng Wang, Maohao Shen).

Conference Reviewing

NeurIPS, ICML, ICLR, AISTATS, ALT, IEEE ISIT, IEEE GLOBECOM, IEEE JSAIT, IEEE Transactions on Information Theory, Mathematics of Operations Research, IEEE Transactions on Wireless Comm., IEEE Comm. Letters.

Community

Organize: Meta, GenAI LLM Compression Reading Group

UCLA EAA Student Projects Judge [Oct. 2025]

Personal Things

Experience

Meta

Research Scientist, Superintelligence Lab

Feb 2024 - Present

RLHF and post-training pipelines for multi-modal Llama models.

UCLA

Graduate Research Assistant

Sample and computationally efficient online learning.

Meta AI

Research Intern

Developed memory efficient techniques for large models inference.

Education

University of California, Los Angeles (UCLA)

Ph.D. in Electrical & Computer Engineering

Dec 2024

Advisors: Prof. Christina Fragouli, Lin F. Yang. GPA: 4.0/4.0.
Distinguished Ph.D. Dissertation Research Award.

Nile University, Egypt

M.Sc. in ECE

Cairo University, Egypt

B.Sc. in Electrical Engineering

Let's Connect

ohanna@ucla.edu