Saurabh Goyal

I specialize in Machine Learning, model compression and optimization, and Generative AI, with applications in Computer Vision and Natural Language Processing. My work includes developing ML models for severely resource-constrained IoT devices with as little as 2 KB of RAM. I have also explored redundancies in ML models to enhance inference time, memory utilization, and storage efficiency for both CNN and Transformer-based architectures. Beyond model compression, I have worked on multi-modal Transformer models for information extraction from business documents, focusing on key-value pair extraction and handwriting recognition. Currently, as part of the AI for Business Automation group, I am leveraging Large Language Models (LLMs) and agentic frameworks to automate complex workflows.