$ hello, world

Builder by default.
Consultant by training.
I find what systems obscure, then build what makes them honest.

I build AI-powered products that solve real operational problems — from enterprise GenAI platforms at McKinsey to open-source developer tools. Based in San Francisco.

Background

Engagement Manager at McKinsey Technology, where I've led GenAI implementations, production optimizations, and operating model redesigns across industries from steel manufacturing to federal agencies. Before that, four years at the Massachusetts Department of Revenue building data infrastructure for $30B tax revenue operations. Two degrees from Bentley University, both earned while working full-time.

Full story and resume

Projects

View all →

Enterprise GenAI Knowledge Platform

McKinsey · 2025

"Strategic outputs shouldn't die the moment the engagement ends."

Built and deployed a GenAI-powered knowledge system across 3 client sites, turning static consulting deliverables into living, queryable intelligence. Sole business lead; scaled to 40–50 person cross-region team.

Founding EM 3 deployments → Dedicated team of 5+ PMs

Project Ontos

Open Source

"AI context should be readable, not retrievable. Glass box, not black box."

A local-first AI context management system: a repo-native knowledge graph in markdown + YAML, designed for inspectable, repeatable workflows across AI tools. Solo-built and published on PyPI.

Python CLI PyPI: ontos 500+ commits Apache 2.0

Folio

Personal

"Your past work should be searchable, not buried in a folder."

CLI tool that transforms PowerPoint and PDF deliverables into a searchable, AI-native knowledge library. Converts decks into markdown with LLM analysis and version tracking.

Python · Click CLI Claude API Obsidian-compatible