Projects
Every system I've built and run.
A mix of automations, data platforms, and production software. Here's what each one does, and what it changed.
Project · Content automation
Multi-Platform Content Automation Pipeline
One draft goes in. A finished article, a set of branded assets, and a scheduled week of platform-native posts come out — published to the blog and across Instagram, LinkedIn, and Pinterest, hands-off after a single approval.
Project · Agent tooling
Reusable Vision & Image Skill for Claude Code
Claude Code can read images and sketch visuals, but only so far — Gemini's vision and image generation go well past it. I built one Gemini-backed skill that puts see and draw behind a single command — a reusable gateway every project, agent, and skill builds on.
Project · AI product photography
Studio-Grade AI Product Photography Pipeline
A few phone photos of a garment go in. A full set of studio-grade shots comes out — flat-lays, on-model, detail crops — in minutes, checked against the real product.
Product · Gym management SaaS
Multi-Tenant Gym Management Platform
A complete gym-management SaaS — members, billing, inventory, payroll, and analytics — built as three apps over one FastAPI service, multi-tenant and trilingual, shipped to production through CI/CD.
Project · RAG backend
Production-Grade Document RAG Backend
The backend for a document-QA assistant: upload PDFs, ask in plain language, get cited answers. An async RAG pipeline — OCR, hybrid retrieval, reranking, generation — spread across worker pools, fully observable, and not locked to any one AI vendor.
R&D · Postgres optimization
Postgres Partitioning for 95% Faster Queries
Two large Postgres databases with drifted schemas, merged into one through a reversible, idempotent load — then the 56-million-row table declaratively partitioned and benchmarked to at least 95% faster on the filters that matter.
Project · Brand systems
Agent-Ready Brand Consistency Skills
Two skills that pin a project's visual identity and voice into a DESIGN.md and a VOICE.md — machine-readable files every agent loads and self-checks against. Author the brand once; stop re-briefing Claude every session.
Have a process like these?
Tell me what you're working on, and I'll come back with whether I can help and what a first step looks like.