Sri Harsha Gouru
Most of what I do starts with a question I can't let go of. Lately it's been what's actually happening inside a language model at inference time — so I built tools to watch it happen. Before that it was can machines pay each other — so I wired up micropayments on Base with x402.
I tend to work across the full surface — training small models, building the infra to serve them, writing the interfaces people interact with. The interesting problems usually live at the seams between these layers.
AI systems, onchain protocols, developer tools — whatever the medium, the through-line is understanding how things work well enough to make new things from them.
Selected work
all projects →LLM Inspector
2026Real-time visualization of LLM internals — trace token generation, attention patterns, hidden states, and probability distributions as they happen. Built to understand what's actually going on inside these models.
x402 Pay-Per-Request API
2026Experimenting with Coinbase's x402 protocol — HTTP-native micropayments on Base using USDC. Server middleware intercepts requests, client auto-signs payments. Exploring what autonomous agent-to-agent commerce looks like.
Tinker Studio
2026Web-based platform for training and fine-tuning LLMs. Experiment tracking, dataset management, real-time loss curves, and an interactive playground to test checkpoints mid-training.