I help businesses build and maintain production web applications — from PWAs and offline-first architectures to backend APIs and CI/CD tooling. Over 20+ years I've worked across scientific R&D, test & measurement, manufacturing, insurance, and e-commerce, so I'm comfortable translating between "what the business needs" and "what the code has to do."

• 🛠️ Current Focus: Progressive Web Apps, offline-first architecture, developer tooling, and AI
• 🤖 I run my own local AI stack — not just an API key: self-hosted LLM service configuration, an MCP server constellation, custom tool-routing prompts for my dev shop, and a persistent dev-container for secure open-source development. I work with agents daily because I operate the infrastructure they run on.
• 🌍 Location: USA (Remote)
• 💼 Services: Full-stack development, technical auditing, team augmentation, automation, AI-stack setup & agent-workflow design

JavaScript

Golang

Node.js

React

Docker

MariaDB

PostgreSQL

Reference architecture for building PWAs with vanilla JS.

• What it demonstrates: Static/dynamic data handling, service-worker-based offline-first design, no framework dependency
• Stack: JavaScript, Service Workers, MariaDB (via companion Nodejs propsdb-api service)

High-performance Go data service, built as a drop-in replacement for the jam-build Node/Express service.

• What it demonstrates: Cross-language service replacement without breaking API compatibility
• Engineering details: Multi-database support (MariaDB, MySQL, PostgreSQL, SQLite, SQL Server) via GORM, Authorizer-based auth integration, optimistic locking with version-conflict detection, tested with testcontainers, shipped as a public Docker image
• Stack: Go, Fiber, GORM, Docker

Deterministic generator that turns a curated, evidence-anchored rule corpus into tiered instructions for coding agents (VS Code Copilot skills and AGENTS.md) — with the tooling to keep those rules honest.

• What it demonstrates: Treating agent instructions as compiled data rather than rotting prose — every rule carries an evidence anchor into the reference repo, so drift surfaces as a named rule ID instead of silent staleness; and measuring real skill value (conformance delta vs. a no-skills baseline) using only the AI stack you already run
• Engineering details: Pure-Node, zero runtime dependencies; deterministic compilation (same input → byte-identical output, snapshot-testable); anchor-drift verifier with durable/volatile severity classes and CI-ready exit codes; tiered output with a hard 20-line always-resident budget; LLM mining as suggestion-only layer; golden-task conformance evals executed in-chat by subagents of your existing AI stack — no new LLM services, API keys, or AI infrastructure required
• Stack: Node.js, Your AI Stack

Build-time library for generating CSP (Content Security Policy) script/style hashes.

• What it demonstrates: Security-conscious tooling — solves a real, easy-to-get-wrong problem (keeping CSP headers in sync with build output)
• Proof: 1500+ weekly npm downloads — actively used by other developers
• Stack: Node.js, integrates into Gulp/build pipelines, published on npm

I am currently accepting select engineering contracts and technical consultation roles — including AI-stack setup, auditing, and agent-workflow design for teams that want to run their own local AI rather than rent it.

• 📧 Direct Email: [email protected]
• 💼 Professional Network: Connect on LinkedIn
• 🌐 Portfolio: localnerve.com
• 🐦 Mastodon: @localnerve
• 🐦 X / Twitter: @localnerve