Work
Real problems. Real outcomes.
Client work
Paid engagements: production systems built and maintained for operator-led businesses.
CSL Capital
Manual risk review. No ML. Reports took 30 minutes to run.
840 API calls reduced to 6. Reports now run in 26 seconds.
Python · BigQuery · ML · GCP · HubSpot
A Residential Solar & Home-Services Operator
No forward visibility into installs. Sales and ops on different numbers.
First accurate 6-month install forecast, on a live executive dashboard.
DOMO · Python · ETL · Forecasting
An Autism-Services Provider
Staff running the same manual workflows 15 minutes at a time, plus a monthly process that consumed a full hour to harden.
Both daily processes cut to <5 min. Monthly hardening down from 1 hr to 20 min.
Process Automation · Python · Workflow Design
Live demos and experiments
Personal R&D: public apps I built to pressure-test tools and ideas. Not client work, but the same engineering discipline.
DroughtWatch
NCAA scoring droughts influence outcomes. No tool tracked them.
305K plays analyzed. 350+ teams. Live during March Madness.
Next.js · Supabase · Python · ESPN API
OC Tendencies
No public tool for scouting OC play-calling tendencies.
113K plays, 44 coaches profiled, used for real game prep.
Python · CFBD API · Supabase · Next.js
Rival Alerts
Rival fans miss the moment their team wins the day.
Multi-sport rival-loss alerts with auto-generated roast copy. CFB + MLB.
Next.js · Neon · Python · Resend
Pickleball Strategy Lab
No tool surfacing real shot-strategy patterns across pickleball skill tiers.
304K shots classified, 935 games, interactive dashboard built in an afternoon.
Python · BigQuery · Gemini Vision · Recharts
Working on something similar?
Start with a diagnostic: a fixed-scope read of your data stack, credited toward the build.
See how I'd approach yours →