Shipped · 2022

NBA Prediction ML API

An XGBoost model achieving 80% accuracy on NBA game predictions, served through a complete ETL pipeline and Flask API.

Overview

ML-powered REST API predicting NBA game outcomes with 80% accuracy using XGBoost. 40+ engineered features from team stats and historical matchups. Complete ETL pipeline and Flask API.

The Problem

Most public NBA prediction models achieve 65-70% accuracy. Getting to 80% requires careful feature engineering.

The Solution

Complete ETL pipeline collecting NBA statistics. 40+ engineered features. Tuned XGBoost classifier. Wrapped in Flask REST API.

Key Results

80% prediction accuracy

40+ engineered features

Complete ETL pipeline

Flask REST API

Tech Stack

PythonXGBoostFlask

Links

Source private

More work

OpenClaw

Multi-agent orchestration platform powering Zenith — a persistent AI chief of staff coordinating a fleet of specialized agents across investment, healthcare, and content workflows

↗

Coach Claw

Open-source coaching layer for Claude Code with local transcript analysis, ambient tips, XP, streaks, and level-ups

↗

AI Companion

Holographic AI companion with custom Jetson camera firmware — open-sourced the IMX708 driver for JetPack 6.2

↗