My mission is to turn data into decisions that make life better — blending the rigour of physics and mathematics with the creativity of AI. I envision a future where intelligent systems warn communities before floods strike, keep businesses one step ahead, and put powerful technology in everyone's hands.
Motto“A dreamer never gives up until the dream comes true.”
A multi-branch early-warning system for Sri Lanka. CNN / U-Net models read satellite imagery, XGBoost and LSTM handle tabular weather data, and SHAP explains each prediction — served through a FastAPI backend with a Next.js dashboard and Flutter mobile app.
An AI-driven inventory system that forecasts product demand to optimise stock levels, supporting balanced purchasing and sales decisions and reducing inventory losses.
A voice and hand-gesture desktop assistant built entirely on free tools. Wakes on "Hey Jarvis", a double clap or a gesture; supports English and Sinhala, a "core plus skills" design and click-only confirmation for risky actions.
Compared Logistic Regression, SVM, XGBoost, ANN and CNN on a Kaggle dataset using accuracy, precision, recall and F1 — selecting XGBoost for its performance and analysing deep-learning limits on small data.
Predicting Home / Draw / Away results of European football matches before kickoff from 2008–2026 data, with time-based splits, feature-set ablation and Macro-F1 evaluation.
An interactive web app that grows out of the prediction project — change match conditions and see how the predicted outcome shifts.
A web-based system for tourist bookings and operations with advance reservations — tour packages, scheduling, customer preferences and boat-owner management.
A full-featured membership system built on core OOP principles — plans, trainers, payments, users and class scheduling (Yoga, Zumba, Boxing), deployed on Apache Tomcat.
An end-to-end DWBI solution on the Brazilian Olist dataset (~99K orders, 112K order items, 9 source tables). Built a SQL Server staging area and T-SQL ETL pipeline into a star schema, a Logistics Data Mart, and a 3-page Power BI dashboard tracking 13.2M in revenue, freight costs and delivery performance — surfacing insights such as SP and RJ driving over half of revenue and deliveries arriving ~15 days ahead of estimates.
An ESP32 smart home security prototype with PIR motion and MQ-2 gas sensors for real-time intrusion and gas-leak detection, controlled remotely through the Blynk app with instant alerts.
University
SEDS SLIIT · Biotechnical Division · Ongoing
SEDS SLIIT
Mathematics Students' Society, University of Kelaniya
School · Royal College
All-Island Oriental Music Competition
All-Island Oriental Music Competition
Oriental Music Society
Music Qualifications
Completed
Completed up to Grade 3
Have a project in mind or looking for an ML developer?
Find me on