Rasa Khosrowshahi

I’m a Computer Science PhD student at Brock University working under supervision of Prof. Shahryar Rahnamayan and Prof. Beatrice Ombuki-Berman.

My research focuses on gradient-free optimization and learning for memory compression and other challenging machine learning problems, especially when gradients are unavailable, expensive, or difficult to use effectively. I use gradient-free evolutionary algorithms with subspace optimization for efficient fine-tuning, working at the intersection of learning, optimization, and model efficiency. More broadly, my interests include multi-task and single-task learning, multi-objective learning, representation learning, machine unlearning, responsible AI, and dimensionality reduction in optimization.

Previously, I completed my MASc. in Computer Engineering at Ontario Tech University , where I worked on center-based sampling, large-scale global optimization, differential evolution, and multi-objective optimization.

I also work on gradient-based multi-objective learning through research software. I build open-source ML infrastructure for scalable experimentation, PyTorch training workflows, and reproducible optimization research, and I am currently contributing to the TorchJD framework, an open-source PyTorch library for Jacobian descent and multi-objective optimization.

Alongside research, I enjoy teaching, mentoring, and course design. I taught Topics in Computational Intelligence (COSC 4P96) at Brock University and have assisted courses in artificial intelligence, data structures, machine learning, engineering design, and software engineering.

Brock students interested in undergraduate research projects, including COSC 3P99 or COSC 4F90, with me and Prof. Beatrice Ombuki-Berman are welcome to reach out.