Igor Molybog

Igor Molybog is an Assistant Professor at the University of Hawai'i at Manoa, holding joint appointments in the Departments of Electrical and Computer Engineering and Information and Computer Sciences. His research focuses on advancing artificial intelligence, particularly through large language models (LLMs), multimodal modeling, and core machine learning optimization. He leads the HawAII research group, exploring applications like LLM alignment, efficient inference systems, and scaling properties of foundation models.

Education: Ph.D. in Engineering from UC Berkeley (2022), specializing in optimization algorithms for complex systems. Previously worked at Meta AI on LLaMa model development.

Research Interests:

• Efficient LLM development and evaluation frameworks
• Multimodal AI integration (video/audio + text)
• Scalable optimization for large models
• Computational efficiency in training/ inference

Recent Work: Presented REAL alignment method (2024), developed long-context scaling techniques (2023), contributed to Llama 2 chat models (2023). Collaborates with organizations like Epoch AI on scaling challenges.

Teaching: Offers courses in AI, machine learning, and optimization across ECE and ICS departments.

Labs/Teams: Leads HawAII Initiative fostering AI collaboration at UH Manoa, organizes paper reading seminars, and hosts technical talks with industry experts.