Mind Games

An IWF-compliant Olympic weightlifting competition simulator, built end-to-end as a data platform on ~7,000 real USAW competition records.

★ anchor🖥 live demo🔒 private repo

Open live demo ↗

The Build Loop

Because I compete in USAW-sanctioned Olympic weightlifting and know the IWF rules cold, I kept hitting the fact that there was nowhere to practice the strategy of a meet — the attempt selection, the clock, the "who's leading after this lift" math, so I built a full competition simulator on real historical data that lets you run a meet the way it actually plays out. It taught me that deep domain expertise is the moat — the rules knowledge is what made the simulation correct, and the correctness is what makes it valuable.

What it looks like

Landing — "Analyze Every Lifter. Simulate Every Meet." on ~7,000 real USAW records.
Full landing page — the data platform pitch coaches use to win Nationals.
Data Browser — searching 68,637 lifters across 7,358 competitions of public USAW results.
Competition Mode — simulate a live meet: attempt selection, the clock, who leads.
Warmup Generator — timed warm-up plan derived from the opener.
Recorded walk-through of the running app.

Mind Games

One-liner: An IWF-compliant Olympic weightlifting competition simulator, built end-to-end as a data platform on ~7,000 real USAW competition records.

Origin (the Build Loop)

Because I compete in USAW-sanctioned Olympic weightlifting and know the IWF rules cold, I kept hitting the fact that there was nowhere to practice the strategy of a meet — the attempt selection, the clock, the “who’s leading after this lift” math, so I built a full competition simulator on real historical data that lets you run a meet the way it actually plays out. It taught me that deep domain expertise is the moat — the rules knowledge is what made the simulation correct, and the correctness is what makes it valuable.

The problem

Olympic weightlifting meets are as much strategy as strength: when you open, how you jump, whether you chase a placing or play it safe, all under a running clock and IWF rules. There’s no flight simulator for that. And the historical data that would make a realistic simulator lives inside USAW’s Sport80 portal — behind a SAML login, a cookie-consent step, and paginated tables, exported as thousands of individual CSVs.

What it does

  • Competition Mode — simulates a sanctioned Olympic weightlifting meet under IWF rules: 2-minute clocks, weight declarations, attempt order, and live standings.
  • Data Browser — searches a live database of 68,637 lifters across 7,358 competitions of public USAW results, per-lifter bests and history.
  • Warmup Generator, Speed Run, Dice, and Masters modes — a suite of practice tools all built on the same historical dataset.
  • Runs on real historical athlete/meet data rather than synthetic numbers.
  • Full web app UI (React + Radix) for running and reviewing simulated meets, live at mindgames.fit.

How it’s built

  • Stack: TypeScript, React, Radix UI, Tailwind; Neon (serverless PostgreSQL); Drizzle ORM; Playwright; Vite; Vitest. Node/Express server, etl/ pipeline, scripts/, migrations/.
  • Notable engineering:
    • ETL at scale — ingests ~7,000 USAW CSV files into Neon PostgreSQL through a repeatable pipeline (not a one-time hand import).
    • Authenticated data collection — custom Playwright automation that logs in through Sport80’s SAML flow, handles the cookie-consent step, and walks paginated result tables to export the full dataset reliably.
    • Rules-correct domain model — the schema and simulation encode IWF competition rules, which is what separates this from a generic sports database.
  • Architecture: Playwright scrapers → normalized CSVs → ETL into Neon PG (Drizzle-managed schema/migrations) → Express API → React simulation client.

Proof points

  • ~7,000 USAW CSV files ingested into a production PostgreSQL database — surfacing 68,637 lifters across 7,358 competitions in the live Data Browser.
  • Automated a real SAML login + cookie-consent + pagination flow to collect the data at scale.
  • Pre-launch hardening complete (Sept 2026) — full-scale audit fixes, TypeScript clean, 215 tests green, zero CRITICAL/HIGH issues remaining.
  • End-to-end: scraper → ETL → database → API → interactive simulator, all one person.
  • Separate USA Masters dataset/config — the platform generalizes beyond one data source.

What to show

  • Demo: Deploy the Replit build to a public URL (repo stays private). Seed with a real historical meet so a visitor can run a simulation in one click.
  • Visuals needed: the simulation screen mid-meet; a standings/leaderboard view; a small architecture diagram (scrapers → ETL → Neon → UI); optionally a screenshot of the ETL run count.

Case-study angle

The clearest proof that Alex builds where his lived expertise is deepest: a rules-correct Olympic-lifting simulator that only exists because he knew the sport well enough to model it — and was persistent enough to build the authenticated pipeline that collected 7,000 files to feed it. This is the anchor case study for the whole portfolio.