Skip to work experience
Matthew Kooy.

University of Michigan / Data Science & Economics / 2027

MatthewKooy.

I build and maintain production software and data systems for real users.

GitHub LinkedIn

01 / Open the toolkit

A few of my
working tools.

Continue to experience ↘

02 / Experience

Work experience.

Software Engineer Intern · HybridMarch 2026 – Present

Team Financial Group

A multi-tenant customer portal for agreements, payments, and documents, backed by guarded Python ETL.

ReactExpressPythonSupabaseMicrosoft GraphCloud Run

What I built

  • Customer portal

    Account-scoped access, invitations and roles, and SharePoint documents with fail-closed authorization.

  • Data pipelines

    Python ETL across LeaseWorks, Salesforce, Supabase, and SharePoint, with validation and manual-review gates.

  • Releases & operations

    Health-gated Cloud Run releases and an operations cockpit for pipeline status, run details, and database diffs.

Sports Operations Intern · RemoteJuly 2026 – Present

AirPLAi Sports

Player and team analytics, video-enabled shot charts, and a reproducible basketball tracking pipeline.

Next.jsTypeScriptPostgreSQLPythonYOLOTrackEval

What I built

  • Player analytics

    Production improvements to responsive player/team profiles and video-enabled shot-chart workflows.

  • Query performance

    Batched PostgreSQL reads replaced per-record query fan-out, with real-database regression coverage.

  • Basketball tracking

    YOLO detection, active-player filtering, identity linking, MOT validation, and TrackEval evaluation.

03 / Campus involvement

Extracurriculars.

University of Michigan

Michigan Blockchain Club

Software Engineer, Development Team

January 2025 — Present

  • Automated smart-contract metric ingestion in Python/pandas, cleaning 25K+ rows into analytics-ready tables.
  • Coordinated Git reviews with three teammates across eight pull requests, keeping changes traceable and outputs reproducible.
  • Python
  • pandas
  • Git
  • Data Engineering

Wolverine Sports Analytics

Project Team Lead

January 2025 — Present

  • Modeled NBA player-stat forecasts with Python regression across 1K+ games for repeatable weekly analysis.
  • Built 20+ matchup and defense features with validation checks and structured feature tables for error tracking.
  • Python
  • Regression
  • Sports Analytics
  • Feature Engineering

IPO Investing Club

Junior Analyst

January 2024 — Present

  • Built Excel reporting packs comparing 12 metrics per company across KPIs, growth, and margins.
  • Standardized valuation templates and assumption logs to keep recurring company coverage consistent.
  • Excel
  • Financial Analysis
  • Valuation
  • Reporting

04 / Projects

Projects.

09 projects

Data Engineering

NBA Analytics Warehouse

A resumable warehouse spanning 30 NBA seasons and 18.3M play-by-play events, built for reproducible basketball analysis.

  • Python
  • DuckDB
  • Parquet
  • Data Quality
  • ETL
Developer Tools

Lewis

A local-first AI engineering assistant with durable memory, isolated Git worktrees, and human-approved code changes.

  • Python
  • FastAPI
  • AI Agents
  • Local-first
  • Developer Tools
  • Approval Workflows
Full-Stack Development

Team Financial Group Customer Portal

Built and deployed a full-stack customer portal for authenticated users to access agreements, payments, documents, and account-specific data.

  • React
  • Vite
  • Supabase
  • Express
  • Microsoft Graph
  • SharePoint
  • Salesforce
Sports Analytics

Fantasy Baseball Daily Briefing

Python pipeline that uses APIs, scraping, and statistical analysis to email daily matchup, injury, waiver, and lineup optimization insights.

  • Python
  • APIs
  • Scraping
  • Statistics
  • Optimization
Machine Learning

NBA Stats Predictor

CatBoost models trained on 10,000+ games from 60 players, improving held-out MAE by 0.62%–1.90% over a 10-game rolling baseline.

  • Python
  • CatBoost
  • FastAPI
  • React
  • SQLite
Code
Systems

Stock Simulator

C++ order-matching engine with O(log n) latency on 100K+ events and 2M ticks/sec market data ingestion.

  • C++
  • Data Structures
  • Performance
  • Algorithms
Systems

SQL Database Emulator

In-memory SQL engine in C++ with parsing, indexing, and test-driven validation for structured queries.

  • C++
  • Databases
  • Parsing
  • Indexing
Machine Learning

Machine Learning Text Classifier

Bernoulli Naive Bayes classifier in C++ trained on 20K+ posts with 25 unit tests for validation.

  • C++
  • Machine Learning
  • NLP
  • Testing
Algorithms

3D Puzzle Solver

Explores 1M+ states in under 2 seconds using BFS and DFS optimization strategies.

  • C++
  • BFS
  • DFS
  • Optimization

05 / NBA Analytics Warehouse

Thirty seasons.
18.3 million events.

Resumable backfills.
Reconciled identities.
Reproducible analysis.

18.3Mplay-by-play events

Python · DuckDB · Parquet

Explore the warehouse ↗