Neil Godfrey
Software engineer. Builder of small useful things. Written down while it happens.
Neil Godfrey is a software engineer building Godfrey Labs — a small portfolio of products, each shipped in the open: paiddaily.io (a read-only desk for tracking market opportunity), sagedaily.io (a tarot and astrology reading agent), eightballsports.com (free squares and strip boards that settle off the real score), elevatedaily.io (a private daily practice), truthers.io (a forensic agent reading the public record), commercialgasketservices.com (a gasket trade and the software that runs it), and Build Daily itself.
The code runs across the stack: Next.js and TypeScript on the frontend, Python and Node on the backend, AWS underneath, and a deliberate split on the AI side — a hosted model where a person is waiting on the answer, local models for the batch work where nobody is, and Neo4j behind the agent memory.
Before this came a decade-plus at an enterprise SaaS serving 50+ corporations and 400+ facilities — two flagship products owned end to end, a maintenance-scheduling system and a utility metering and energy platform, each from schema through web, mobile, and deployment, plus the modernization of a ten-year-old .NET codebase onto .NET Core and Next.js. Long-lived software for people whose day stops when it breaks is a different discipline from greenfield, and most of what shows up here was learned there.
What this site is
Build Daily is the public surface of that work. Every post, daily log, and project page here is a real artifact from an active build — not content written to rank, but notes written because the work happened and writing it down makes it sharper. The daily logs include the days things broke, which are usually the ones worth reading.
The site is statically exported from Next.js and served through CloudFront, with the live surfaces on Lambda behind it. The agent on the homepage answers by retrieving from this published corpus rather than being trained on it — it can only say what the site already says, which is the point.
The stack in use
- Frontend: Next.js, React, TypeScript, Tailwind
- Backend: Python, Node.js, FastAPI
- AI / Agents: DSPy for typed LLM programs, Ollama for local inference, a hosted provider where latency matters, Claude Code, Neo4j-backed agent memory, retrieval pipelines
- Infrastructure: AWS (S3, CloudFront, Lambda), Docker, Neo4j, PostgreSQL, a self-hosted GPU box
- Earlier work also covers: mobile (iOS/Android), .NET → Next.js modernization, on-chain integrations (Base, Ethereum)