Qi Heng Leow

I've built end-to-end forecasting and prediction systems, from data ingestion and model training to cloud deployment and performance analysis. My projects use Python, SQL, machine learning, APIs, SQLite, and Power BI to turn large datasets into actionable forecasts.

For a weather forecasting bot, I blended nine forecast models, ran the pipeline continuously on Oracle Cloud, and analysed forecast error, win rate, and cumulative P&L by city. For a cryptocurrency prediction bot, I processed 45,000 order-book tick files and trained gradient-boosting models across 2.85 million rows.

During my actuarial internship at Prudential BSN Takaful, I validated Excel macros, calculated product usage, cost, and loss-ratio metrics, and reconciled fees during UAT. I've also built and deployed an insurance risk modeling service using FastAPI, Streamlit, Docker, GitHub Actions, and Hugging Face Spaces.