apps · 2026
Corporate Ops Suite
The software stack I build for companies — warehouse platform, engineering calculators, forecasting, and an AI email agent. Shown here as a staged demo.
This is corporate software development as a service: a full operations stack designed, built, and shipped end-to-end by one person for a working manufacturer, running on their floor every day. At its heart is a warehouse platform — a parts catalog four levels deep, a mobile count workflow that turns a 250-tap stock count into typing numbers as you walk the shelves, and nested build-recipes that deduct every component the moment a product package ships. Around it sit the automation tools that run the rest of the business: an engineering calculator that does the real safety-critical spring math, a quoting tool that prices any product configuration in one click, a reorder tracker that cracks package sales back into true part demand, and a Python email connector that hands the company inbox to an AI assistant. It runs local-first so the warehouse never stops when the internet does, reaches a second site over a secure tunnel, and backs itself up offsite every hour. The client and their data stay private — everything shown here is a staged demo build.
- Warehouse platform — four-level parts catalog, multi-token search, mobile inline stock counts with one-tap batch commits, timesheets, role-based access, installable PWA
- Nested build-recipes: one door package expands into every panel, spring, and bracket in a single atomic, audited transaction
- Engineering calculator with real safety-critical spring math — cross-checked against industry tools until the numbers matched every time
- Forecasting that cracks package sales into true part demand, with per-part per-month reorder points tuned to supplier lead times
- AI email agent — a Python FastMCP connector that lets an assistant search, triage, and draft the company inbox
- Customer-facing product visualizer — upload a photo of your home, preview the product on it before you buy
- React
- TypeScript
- Node
- Prisma
- SQLite
- Python
- FastMCP
The forecasting model figured out that one product sale quietly consumes four to five components — a fact the point-of-sale system never once mentioned.
