FerroDB
A relational database built from scratch in Rust — now AI-native with an HNSW vector index
Open to full-time work as a Forward-Deployed Engineer or Applied AI Engineer. I embed with a team, learn how they actually work, and ship the system that replaces the manual process.
Available for contract work →Money tracker for cash workers: shift logs, receipt OCR, tax math. Live.
The work a hiring manager can evaluate directly: read the code, run the live products.
A relational database built from scratch in Rust — now AI-native with an HNSW vector index
Ski & Snowboard Rental Management
Money Tracker for Cash Workers — receipt OCR on Anthropic vision
Casino Strategy Trainer
Ski & snowboard rental shop · Beaverton, OR
I embedded with PTO Ski & Snowboarding's rental-floor workflow and replaced paper forms, manual DIN charts, and four separate tools with one system in daily use. This is the job a Forward-Deployed Engineer does.
Six years in operations management before I wrote software, leading a 12-person team in cannabis, one of the most heavily regulated industries there is. Seed-to-sale tracking, state compliance, and audit trails were the daily job. That is why I can walk into a client's operation and read what they actually need, versus what they said they need. When I later built CannaTrack, a compliance system for that same industry, I was building for a floor I had already run. Forward-deployed engineering is that job: embed, learn the constraints, and remove the manual process.
How I think about building with AI: the trade-offs, not the hype.
“We went from paper forms and manual DIN calculations to a complete digital rental operation. Check-ins are faster, our inventory is tracked in real time, and the whole team was up and running on day one. PowderLedger runs our entire shop now.”
I own the whole path: architecture, implementation, evals, and deploy. I write the spec before the code and document the decisions the model could not make.
Before I write code, I document the features, the data model, and the edge cases, and we agree on the spec. The build matches the spec, and the decisions get written down so the next person inherits the reasoning, not just the diff.
I take a system from schema to production: auth, data model, the frontend, and the deploy. I care as much about the parts that are not glamorous, the eval harness, the failure analysis, and the fallbacks, as the demo.
Six years running operations taught me to scope from how the work actually happens, not from a feature list handed down. I ask what breaks on the floor before I build, which is the core of forward-deployed work.
8production applications shipped, 6 in daily use. Systems people run, not demos.
SaaS platforms, LLM-integrated products, PWAs, fintech, dashboards, and APIs. From a single AI feature to a full production system.
Define the problem, map the user journey, design the database schema, plan the API surface. Every project starts with a clear architecture before a single line of code.
Claude Code as a pair programmer. Master prompts for specification. Multi-agent workflows for implementation, with verification gates that keep the output honest. AI handles the repetitive work so I can spend my time on architecture, evals, and the decisions an AI can’t make.
Push to production on Vercel with CI/CD. Every commit auto-deploys. Real users, real feedback, rapid iteration. A product isn't done when it's built. It's done when it's serving people.
Full-time is the goal, but I also take on contract builds. The protections and services below are for contract work.
Four clear ways I can help, from a full SaaS build to an architecture review. Milestone-based delivery means you see real progress at every step.
Full-time as a Forward-Deployed or Applied AI Engineer is the goal. If that is the role you are filling, get in touch. Open to contract work too.
Forward-Deployed Engineer · Applied AI Engineer · Full-time or contract