A working PI needed his casework — documents, interviews, reports, email — in one secure place. I designed and shipped the entire platform: native macOS, iPhone + Apple Watch, and a full web app, all in sync through a backend on my own infrastructure.
VINdex turns a vehicle identification number into verified compatible parts — a deterministic fitment engine, eBay API integration, and automated browser sourcing across suppliers, all self-hosted and cached for speed.
For Apple Cleaning Systems: a full public site with online scheduling — plus Rosie, an AI estimator in production. Customers upload a photo of the space; Rosie identifies the service and returns a price range, instantly. Behind the scenes: internal workflow, operations and deploy apps I run for them.
The reason my AI team never forgets a decision: a knowledge brain with vector search, a self-wiring graph, and nightly consolidation — served through my own MCP server with OAuth, around the clock. This isn’t “using AI.” This is building it.
I built my own private school: a web app with step-by-step playbooks, a sandboxed fake terminal to practice commands safely, and an AI tutor that checks your work. I use it to master my own tools — and the same system can train a team, a company, or students on any new software. Login-only · by design.
My own native desktop app (Mac, portable to Windows) where every little tool I need becomes a module: an AI document converter, a dictation cleaner, file encryption, automated Drive backups, a SoundCloud publisher, a project dashboard. Seven modules and counting — proof that even “small personal tools” deserve real software.
One-command notarized releases, automated daily disaster-recovery backups, code knowledge-graphs across 12 repositories, and a deploy system for my own VPS. The quiet machinery that lets one person run all of the above.
A friend once spent months not asking me for something he needed badly — converting iPhone photos to JPEG in bulk on his Windows PC — because in his head it sounded too hard, too expensive, too much to ask. When he finally did, my AI team and I built it, gave it a clean interface, packaged it as a Windows app, and installed it on his machine… the same day.
People think every piece of software is a mountain. Some are — you’ve seen mine above. But many of the tools that would change your workday are hills we can climb in an afternoon. So bring me the idea, however small or however wild. We’ll figure it out, and we’ll get it done.
The theme engine, the VIN decoder and the AI chat below run for real, right now — rate-limited, because I engineer AI costs like everything else.
Type any 17-character VIN, get the vehicle — the first step of the VINdex pipeline, decoded live.
Ask anything about my work — answered live by the same agents that build with me.
Hear the ReyPI pipeline: a scanned document, translated, read in a natural voice.
Claude, OpenAI and Gemini models orchestrated as a real team — with their own long-term memory, code maps of every repo, and shared debugging lessons. I direct; they multiply.
Design → build → automated verification with real browsers and simulators → deploy to my own servers → over-the-air updates. Every step scripted. Every release repeatable.
AI writes fast. I make it correct, designed, and shipped — with the detail work that separates a premium product from a template.
iPhone, Mac, web — or all of them in sync. From idea to real distribution: App Store, notarized installers, or production web. The way ReyPI was built.
Voice, transcription, document intelligence, search, agents — added to what you already have, with API costs engineered down, the way I run my own.
Your app lives on managed infrastructure with monitoring, updates, and a human who answers. The quiet reason everything on this page stays up.
Contract or full-time, remote, English or Spanish. Tell me what you need — you’ve just seen what I ship.