David Hardage

Senior Analytics Engineer in Texas. I build data visualizations and write about tech, economics, and data.

Featured Project

cinemetrics

Eight years of film-watching, end to end

A full analytics pipeline over my own Letterboxd history. I enrich the ratings from the TMDB and OMDb APIs, model them in DuckDB with dbt, and serve the result as a static cross-filtered dashboard with an ML recommendation engine. Every stage runs in a CI runner or in your browser, avoiding the need for an application server or a hosted database.

  • Every chart responds to every filter in real time and every view generates a shareable URL.
  • Guided stories surface a finding in one tap. Each one filters the charts and annotates what they show.
  • The recommendation engine ranks candidates by cosine similarity to a taste vector built from my ratings, then explains why each one surfaced.
  • A GitHub Action pulls new ratings every day, retrains the embeddings, and redeploys.
PythondbtDuckDBscikit-learnNext.jsD3Cloudflare R2

More Projects

Capital Flow Cityscape

A 3D cityscape where buildings represent financial securities and capital flows become visible paths through a living city. Unique city every page load.

Three.jsProcedural Generation

UrbanStack

Multi-layer urban data platform combining transportation, infrastructure, walkability, and demographic data for major US cities.

Next.jsdeck.glDuckDB

World Cup 2026

Interactive map exploring where every player was born versus which national team they represent.

Next.jsdeck.glMapLibre

Coach's Challenge

NFL coaching decision quality analyzer — grades play-calling against 27 years of EPA-optimal baselines.

Next.jsRechartsDuckDB