Projects

The six-year story of helping grow the Child Poverty Action Lab's data function: hiring the org's first data engineer, pairing vendor engineers with an in-house team to build the platform, selecting and migrating to Databricks, leading the data org through a CDO transition, and rolling out AI-enabled workflows across the team. A leadership case study, not a pipeline write-up.

• > Helped grow the team from 1 analyst to 2 in-house + 4 vendor engineers
• > Led the data org through a CDO transition
• > Selected and migrated to Databricks; shaped the org-wide Claude Enterprise rollout

The eviction data workstream behind the Child Poverty Action Lab's housing work: turning messy county court records into a daily feed that 12+ partners (the Dallas Eviction Advocacy Center, the Princeton Eviction Lab, Dallas HHS) use to reach tenants facing displacement, and into the public transparency of North Texas Evictions. Built in R in 2020, then rebuilt on Databricks by our data engineering team as it scaled to four North Texas counties. What mattered was always who the data reaches.

• > ~48,000 Dallas filings a year turned into tenant outreach
• > ~800 tenants a year connected to legal representation
• > Feeds 12+ partners and North Texas Evictions

Interactive R Shiny application mapping homestead exemption take-up rates across Dallas by census tract, zip code, and neighborhood. Filters by race/ethnicity and income level to surface disparities in who is claiming property tax savings. Built with Mapbox GL, sourcing data from the Dallas Central Appraisal District and American Community Survey.

• > Choropleth mapping by tract, zip, and neighborhood
• > Equity-focused filters by race/ethnicity and income
• > Downloadable report and open data

The on-the-ground companion to the Homestead Exemption Map. Around a canvasser's location, it flags every parcel likely eligible for an unclaimed homestead exemption (worth hundreds to thousands of dollars a year per household), estimates the tax savings left on the table, and tracks outreach status door to door. Built with R Shiny and Mapbox GL on Dallas Central Appraisal District data, it's used by roughly 20 field-team members across CPAL's outreach partners.

• > Flags homes likely missing a homestead exemption, with projected savings
• > ~20 active field-team users across outreach partners
• > The door-to-door companion to the Homestead Exemption Map

A community-sourced tournament data platform for the Digimon and Gundam card game scenes. A 600-member Discord and a network of volunteer organizers keep the data flowing; the platform turns it into regional meta analysis, scene-aware leaderboards, and blog deep-dives players use to prep for events. So far: 10,800+ tournaments logged, 19,100+ players tracked, and around 2,000 daily active users. A public API powers external community tools. Built on Astro, Preact, and Neon Postgres.

• > 10,800+ tournaments logged · 600 Discord community members
• > SSR Astro + Preact islands · Neon Postgres on Vercel edge
• > Public token-authenticated API and admin CRUD subdomain