What I'm Doing Now

Work

Still at Accenture — currently deep in a Medicare Advantage HRA data platform project on Snowflake. The work involves a lot of data flow design and translating messy healthcare requirements into clean pipeline logic. It's good systems work.

Also actively looking for my next move. Nine years in healthcare tech has given me strong opinions about what good product work looks like, and I want to find somewhere those opinions are actually useful.

Building

Riff Notebook — the guitar practice intelligence project. The core problem I'm trying to solve is that music information retrieval (MIR) completely breaks down for distorted guitar. CREPE degrades, chord recognition degrades, standard tools fail. I'm working through what's actually possible given that constraint, and what a training data strategy looks like from scratch.

Ask the Ancients — a PWA that lets you consult historical philosophers through grounded AI. Still in planning. The design challenge is making it feel like a conversation rather than a search engine.

This site — just launched. I built the whole thing in one session. Now comes the harder part: actually writing.

Reading

Working through a mix of philosophy and technical material. The Stoics, mostly. Marcus Aurelius is evergreen. On the technical side, a lot of MIR papers and audio ML research — trying to understand the state of the art before deciding what's worth building.

Guitar

Playing more than building right now, which is unusual. Drop tunings, heavy stuff. I follow a surprisingly large community of international luthiers on Instagram — one of the few rabbit holes I don't regret. The dream of building my own guitar line is still alive. Name TBD.

Thinking About

• Why most AI products feel like features, not products
• Whether MIR for distorted guitar is an unsolvable problem or just an unfashionable one
• What it means to build a system that outlasts the session that created it
• Agent memory — how do you give an AI genuine continuity across conversations?