Nishant Thakur makes LLMs run locally. As an AI engineer at Adobe, he builds browser-native inference pipelines using WebGPU, ONNX, and WebAssembly, completely bypassing the need for cloud dependencies.
His work focuses on removing friction from AI development. He researches ways to drastically cut LLM tool-calling token overhead and builds entirely local workflows, like PDF-to-podcast pipelines on Apple Silicon. He also teaches developers how to build zero-cost, in-browser AI agents from scratch without requiring API keys or external servers.
Based in Bengaluru, Nishant is a regular speaker at local MLX and AI meetups.