I'm an 18-year-old data scientist studying at the University of Hertfordshire, graduating in November 2026 with a B.Sc. in Data Science. I build end-to-end ML systems that are explainable, deployable, and business-focused — from feature engineering to SHAP insights to production-ready APIs.
What I do:
- 🧪 Feature Engineering — Turn raw, messy data into predictive signals
- 📊 Data Viz & Explainability — Uncover hidden patterns & demystify models with SHAP
- ⚡ Scalable ML — XGBoost & LightGBM for high-performance tabular data
- 🚀 MLOps-lite — FastAPI for real-time predictions + Streamlit for interactive dashboards
- 🔔 Production Alerts — Discord webhooks for real-time fraud alerts
Here's what I've been building — each project is a complete story from raw data to working app.
| Project | What I Built | Key Insight | Stack |
|---|---|---|---|
| E‑Commerce Churn Prediction | Predict if a customer will ever make a second purchase. Live Streamlit dashboard with SHAP explanations. | Delivery delay, not price, is the #1 driver of churn. | LightGBM, Optuna, SHAP, Streamlit |
| Fraud Detection API | Real-time credit card fraud API. Catches 72% of fraud with 50% precision. FastAPI + Discord alerts. | Imbalanced classification works better with optimized thresholds & features. | XGBoost, FastAPI, Discord Webhooks, Streamlit |
| CPU Benchmark Predictor | Predict PassMark score from CPU specs. XGBoost model with R² = 0.983. | threadMark & cores matter more than clock speed or price. |
XGBoost, SHAP, Streamlit |
- Becoming a better data scientist. Learning new skills. Building things that work. Every single day. 🚀
⭐ From raw data to real decisions — explainable, deployable, impactful.