Sai Koushik

My work focuses on the intersection of LLM deployment infrastructure, model compression, and high-performance algorithmic design. I specialize in designing fine-tuning pipelines and building real-time multimodal voice agentic systems and inference pipelines.

I was a Research Assistant at the Penn Computer Assisted Surgery and Outcomes Laboratory, where I built Vision Transformer architectures for identifying surgical phases in robotic procedures. I hold an M.S. in Electrical Engineering from the University of Pennsylvania.

Recent write-up: What Happens to an EKF During a Sensor Blackout?

Selected Writing