Avishkar Saha

I'm a Research Engineer at MireloAI, where I design and train foundation models for video, music, and sound. I completed my PhD in Computer Vision at CVSSP, University of Surrey, supervised by Richard Bowden, Chris Russell and Oscar Mendez. I also interned at Amazon AWS' Causal Representation Learning Lab in Tübingen.

My work at MireloAI involves the complete machine learning pipeline — from large-scale data collection and curation through model architecture design to distributed training of multimodal foundation models. I'm particularly interested in building models that can understand and generate content across multiple modalities.

Google Scholar/Github/LinkedIn/CV