Selected work

Data Journalism

Original investigations built directly on the data warehouse — shell-company networks, mass-litigation, and the businesses most exposed to a downturn.

Fifteen investigative data-journalism stories built on a reproducible pipeline — Databricks → DuckDB → Marimo → publication — with a GREEN/YELLOW/RED bulletproofing protocol that caught defamatory and factual errors before they shipped. Reporting drawn straight from the warehouse, not press releases.

Surfaced 8,747 federal mass-defendant cases from the RECAP archive, and reproduced the NYC Public Advocate’s worst-landlords counts to within 15% from open data alone.

• Databricks
• DuckDB
• Marimo
• RECAP
• Polars

Content-Engineering Platform

An internal platform the engineering org runs on: Claude Skills for the entity graph, a GraphQL identity client, an LLM-evaluation MCP server, and an autonomous agent fleet.

An internal platform the engineering organization runs on: a Claude Code plugin marketplace of 16 production Claude Skills for the Databricks entity-resolution graph plus session tooling, a reusable GraphQL identity-resolution client, and an LLM-evaluation MCP server spanning 14 models across 7 providers.

• Claude Skills
• MCP servers
• GraphQL
• LLM evaluation

Autonomous Content Fleet

Multi-session AI agent loops on systemd and cloud scheduling, running competitive scans, market briefs, and editorial cycles on a cadence.

Multi-session AI agent loops on systemd and cloud scheduling, with monitoring dashboards, running competitive scans, market briefs, and editorial cycles on a cadence — designed, operated, and migrated from a laptop to an always-on VPS substrate.

• Agent orchestration
• systemd
• Monitoring
• VPS

Launch & Brand Systems

Owned the content stack for major launches and partnership announcements — plus the brand-voice system and SEO knowledge base behind them.

Owned the content stack for the Gov Archive launch (1,000+ validated datasets; launch blog through 7+ executive revision rounds; press release and documentation) and the Google Cloud and Clay partnership announcements, including multi-round CEO quote and voice calibration.

Authored the company brand-voice and editorial system; grew an SEO-driven KYB knowledge base to 80+ articles and a 50-term glossary with automated freshness and regulatory maintenance; and stood up a monthly GTM competitive-intelligence program tracking seven identity vendors.

• Launch comms
• Brand voice
• SEO knowledge base
• Competitive intel

Cryptographic Amnesia

A cultural history of how a set of mathematical techniques became an instrument of war, a civil-liberties cause, and a piece of everyone's infrastructure.

A cultural history of how a set of mathematical techniques became an instrument of war, a civil-liberties cause, and a piece of everyone’s infrastructure. Built on archival research across eight sites and interviews with cryptographers and cypherpunks.

• MIT Press
• Archival research
• Oral history

Research & Visualization

Peer-reviewed work on data sensemaking and dirty data; the first study of SecureDrop; the Knight report that shaped curricula at 116 programs.

Peer-reviewed work on data sensemaking and dirty data with Tamara Munzner (ACM CHI, IEEE VIS); the first published study of the SecureDrop whistleblowing platform (Tow Center); and the Knight Foundation report that shaped data-journalism curricula at 116 programs.

• ACM CHI
• IEEE VIS
• Tow Center
• Knight Foundation