Adil Asif

Giving What We Can

I'm a deep learning engineer at NVIDIA, working on NeMo and Megatron-Core, two open-source, large-scale generative AI frameworks. My current research interests focus on reasoning safety and alignment science, with an emphasis on novel methods that scale well. I also collaborate closely with Roger Grosse's group at the University of Toronto.

Until recently, I worked as an Applied ML Specialist at the Vector Institute. Before that, I earned my Honours BSc in Computer Science from the University of Toronto.

If any of this resonates with you and you'd like to chat, please drop me an email.

Experience

Education

Publications

ICLR 2025 & NeurIPS (FITML Workshop) 2024

Teaching LLMs How to Learn with Contextual Fine-Tuning

Younwoo Choi*, Muhammad Adil Asif*, Ziwen Han, John Willes, Rahul G. Krishnan

PaperCode

Contextual Fine-Tuning blends prompting with training to improve how LLMs learn domain-specific knowledge.

NeurIPS SoLaR Workshop 2023

💡 Spotlight Award

FlexModel: A Framework for Interpretability of Distributed Large Language Models

Matthew Choi, Muhammad Adil Asif, John Willes, David B. Emerson

PaperCode

FlexModel provides interpretability tooling for distributed LLMs.

CoRL 2023

Geometry Matching for Multi-Embodiment Grasping

Maria Attarian, Muhammad Adil Asif, Jingzhou Liu, Ruthrash Hari, Animesh Garg, Igor Gilitschenski, Jonathan Tompson

PaperCode

GeoMatch learns contact point likelihood maps and conditional autoregressive predictions of grasps.