Nipun Weerasinghe

A research project developing an OCR system to recognize Sinhala handwritten official documents and extract editable Unicode text. Addresses the gap of no existing Sinhala handwritten OCR covering all characters in fewer than 300 classes. Features include custom dataset creation, image preprocessing, CNN-based character recognition with LLM post-processing, and segmentation algorithms for line and word extraction.

Automated penetration testing tool that scans ports on a target IP and provides detailed summaries with exploit recommendations. Features DOS/DDOS attack protection by tracking traffic and blocking malicious IPs, YARA-based malware scanning, and web server vulnerability detection. Built as a Level 4 university group project where I served as Team Lead & Developer.

An AI-powered quiz generator where users enter a topic, choose difficulty, and select the number of questions. The app generates quizzes with single or multiple correct answers using OpenAI. Incorrect answers reduce the score, and results include correct answers, explanations, and total marks. Version 2 supports essay-type questions.

A real-time dashboard for monitoring and managing ML model deployments, featuring automated performance tracking and drift detection.

An end-to-end natural language processing pipeline for multi-class text classification with transformer-based models.

A scalable ETL pipeline framework for processing and transforming large datasets with automated scheduling and monitoring.