AI researcher and undergraduate at MBZUAI, Abu Dhabi.
Interested in:
LLMs · AI Safety · World Models · Model Evaluation · Interpretability
-
AdaptFM @ ICML 2026 — 6th place, 5.366× speedup on Qwen3.5-4B · Project
-
EdgeRAG @ AAAI 2026 EGSAI Workshop — offline RAG for memory-constrained devices
-
Personal World Models — studying temporal memory, routines, preferences, and changing environments in vision-language models
Exploring how AI models represent information, how their internal behaviour changes, and how we can understand and evaluate them more reliably.