Ali Lesani

Hi, I'm Ali!

I am an AI Scientist at Kisoji Biotechnology, where I develop AI models to support drug discovery research. My work focuses on multimodal antibody representation learning, integrating sequence information from protein language models with structure-based representations from GNN-based models to build richer antibody representations. These representations directly support downstream tasks such as antibody-antigen binding prediction. By learning more informative antibody representations, my work helps filter artificially generated antibody candidates and cluster antibodies with similar properties, supporting more efficient AI-driven antibody discovery.

I am also a PhD researcher at the CViSS Lab, University of Waterloo, supervised by Dr. Chul Min Yeum. My research focuses on interpretable visual understanding, spatio-temporal change detection, and multimodal reasoning via vision–language models. I am developing AI systems that analyze infrastructure inspection images taken over time to automatically detect and describe meaningful changes. The system generates concise change reports along with visual evidence highlighting the regions responsible for those changes, making the outputs interpretable and actionable for inspection workflows.

During my PhD, I have collaborated with industry partners including Rogers Communications, RBC Royal Bank, and NAV Canada through the Mitacs internship program. In these collaborations, I have developed applied AI solutions for real-world engineering and operational challenges, bridging academic research with deployable machine learning systems.

Outside of research, I enjoy playing soccer recreationally and following professional football. I’m also a long-time fan of snooker. If we share similar professional or personal interests, feel free to connect — I’d be glad to chat.