Walt Burge.
AI Developer — Data Science & AI Learning Systems
Self-taught AI developer with a heavy data-science and learning-systems focus. I wrote my first line of code seven months ago; since then I've trained a custom LLM end to end — data curation, synthetic distillation, fine-tuning, and serverless deploy — built the tooling and agent systems around it, and shipped production systems that put real AI in front of real users. I build the model and everything around it: when the thing I needed didn't exist, I built it.
Skills
Selected Work
A custom fine-tuned LLM in production, not a notebook — the whole pipeline: data curation, synthetic distillation, hyperparameter tuning, and serverless deploy at ~$0.002/estimate.
Python · PyTorch · Llama · RunPod
A desktop trainer for the whole fine-tuning loop — GPU provisioning on vast.ai, dataset and hyperparameter management, launch and monitor. The training toolchain, productized.
TypeScript · Next.js · Node.js · Python
A markdown-native language for AI agents — substrate-typed, executed on PyTorch / LangChain / Ollama backends, and formally verified before it runs.
Python · PyTorch · Ollama · LangChain · Markdown
MIT-licensed, local-first AI memory — on-device OCR + an MCP server so any agent can search it. One pip install, nothing leaves the machine.
Python · SQLite · OCR · MCP
AI shipped to real users — a two-sided marketplace where ConstructionAI prices jobs live, across Next.js web, a Java/Spring backend, and a React Native app.
Next.js · TypeScript · Java · Spring Boot · React Native · PostgreSQL
Background
Self-taught AI / ML development, with the full-stack to ship it. Trained a custom model and built the data, training, and agent systems around it — live in production for a paying contractor (MHP Construction, Oxford MS).
BeforeConstruction — ran crews and wrote estimates by hand. That's where the engineering instinct comes from: scope it, sequence it, ship something that holds under load.