Aarash Feizi

I’m passionate about multimodal learning and large multimodal models, exploring their applications (and limitations) in real-world scenarios. Always open to collaborating with like-minded researchers, so don’t feel shy to contact me!

I am majoring in Computer Science at McGill University and Mila, under the supervision of Professor Reihaneh Rabbany and Professor Adriana Romero-Soriano.

My research interests are multi-modal (image/language/graph) representation learning and self-supervision in general.

I recently started as a Student Researcher at Google Research, where I’m working on projects involving text-to-image models and data attribution.

Previously, I was a visiting researcher at ServiceNow Research, where I designed novel multimodal large language models (MLLMs) and explored how open- and closed-source MLLMs perform as reliable judges. Before that, I interned at Recursion, working on multimodal learning for biological data.

Furthermore, I am also interested in how self-supervised methods can enhance their robustness in uni- or multi-modal learning, enabling them to perform effectively across a broader spectrum of applications and real-world datasets.

Finally, I’m the founder of Mila’s Computer Vision Reading Group (currently on pause). If you’d like to attend or present something, send me an email! :)