I take products from ambiguous beginnings to production systems people depend on.
I am a multidisciplinary technologist who moves from vision to architecture to production. I design complex systems, build intelligent products, shape intuitive experiences, and create the technical foundations that allow them to scale. My work spans AI, full-stack engineering, cloud architecture, data, product strategy, and UI/UX, giving me the range to see the whole system and the technical depth to build it.
Product work
Recent product architecture and hands-on builds, from a production insurance chatbot to a functional voice-AI prototype.
Confidential Insurance Platform
An AI chatbot for insurance discovery, rebuilt around explicit state, typed contracts, and modular workflow boundaries.
Contributor — Microsoft Entra ID, AI chatbot, conversation-engine architecture
- TypeScript
- React
- Vite
- Azure
Veradic
Hobby project; Voice AI for freight brokers: cover more loads, make fewer calls.
Solo builder: product, strategy, design, engineering
- TypeScript
- Remix.run v3
- Twilio ConversationRelay
- Cloudflare Durable Objects
Selected work
AI Call Analytics: Natural-Language Answers to 'Why Do Customers Call?'
Sole architect of a production AI pipeline that transcribed and classified about 10,000 calls per day for a multi-billion-dollar rental company. Its findings led to a product change that cut total call volume over 30%.
- ~10,000 calls/day ingested and classified
- Classification: hours to minutes with Batch Inference
- ~30% reduction in total call volume
- First production AI system at the client
Testing Against Production in Seconds: Module-Federation Override Tooling
Built a Webpack 5 runtime-override system and browser extension for a 60-micro-frontend platform whose lower environments did not match production. Developers could test local or branch builds against production conditions in seconds.
- 60 micro-frontends, 4 teams
- Production-parity testing: hours to seconds
- One-click dependency-aware deployments
The $1.4M/Year Bill Nobody Was Watching
A retailer's RFID loss-prevention POC emitted mostly false positives and cost about $119K/month on GCP. I replaced per-second label diffs with windowed SKU-set comparisons, modeled the cloud cost, and cut the bill by about 99%.
- ~$119K/month to ~$1.3K/month (~99% cut)
- ~10 valid events per 10,000 sent, before the fix
- $500–700M ROI opportunity sized in proposal
How I work
Start anywhere
User research, data modeling, frontend, backend, or infra — the work starts wherever the problem starts.
Ship, then keep it healthy
Monitoring, observability, and root-cause debugging are part of the product, not an afterthought.
Any scale
Solo founding engineer, embedded in a team, or leading one — hands-on at every scale.