Adnan Qidwai

u/adnanqidwai/about.md

Hi! I'm Adnan. I currently work at IBM Research as a Research Engineer in the Data for AI team. My work is centered on semantic data pipelines and streamlining how models ingest and utilize large amounts of knowledge.

Beyond my core role, I'm interested in building long-horizon agents, with a particular focus on curating the right training data and creating meaningful metrics to measure their progress. I also enjoy probing frontier models to understand their limits and identifying the specific gaps where their reasoning or reliability starts to drop off.

In the past, I worked on multimodal understanding in collaboration with Prof. Dan Roth (UPenn), Prof. Vivek Gupta (ASU), and Adobe Research. I also spent time researching LLM jailbreaking as part of the Stanford AI Researcher project.

I'm always open to discussing research, new opportunities, or just exchanging ideas. Feel free to reach out!