I help small teams turn data and probabilistic models into products people use.
I’m currently building askdata.app – an LLM-powered interface for exploring structured data in plain English. Before that: product data science at Lyft, and consulting work helping organizations find the metrics that actually matter.
Things I’ve done
- Detected severe patient events 12 seconds earlier than the industry standard, at the same accuracy. Time-series work on continuous patient monitoring data for Bernoulli One – enough lead time to change clinical action and reduce alarm fatigue.
- Shifted a national nonprofit’s North Star from “programs delivered” to student outcomes. Worked with BUILD via Delta Analytics, and built the automated reporting that supported the shift.
- Ran experimentation and data modeling on Lyft Integrity and Delivery – where I learned to focus on users and the why behind what we build.
- Shipping with LLMs – evaluation, prompt patterns, and the long gap between demo and product. Building askdata.app now.
The arc
Data science → product → building. I’m most useful at the seam: the first hire who can do the first data work, ship the first product, and tell you whether it actually moved the metric.
Let’s talk
If you’re early-stage and building with AI, I’d love to talk. hello@kpan.dev.