My personal portfolio

About Me

I am a committed Computer Engineering graduate (with Second Class Upper Divison) specializing in Data Management, with a strong interest in Software and Full-Stack Development, Computer Vision, and Machine Learning. My academic experience has provided a comprehensive understanding of fundamental computer science principles, artificial intelligence, optimization, and mathematics. I am driven by intellectual curiosity and a passion for innovation, continually seeking to leverage emerging technologies to develop efficient and creative solutions to real-world challenges.

Second Class Upper division

GPA: 3.4

5+

Certifications

3+

Major Projects

Education
Certifications

B.Sc Engineering (Hons) - Computer Engineering

University of Sri Jayewardenepura

Specializing in Data Management

2021 - Present CGPA: 3.65

G.C.E Advanced Level (Physical Science)

Maliyadeva College

2019 1A & 2Bs

G.C.E Ordinary Level

Maliyadeva College

2016 9As

Fundamentals of MLOps

KodeKloud, Raghunandana Sanur.

LLM Engineering: Master AI, Large Language Models & Agents

Udemy.com

Programming in Python

open.uom.lk

LangChain Fundamentals

KodeKloud, Janakiram MSV

TensorFlow Developer Certificate Bootcamp

Zero To Mastery (Following)

Machine Learning in Production

DeepLearning.AI on Coursera (Following)

Experience

Software Engineering Intern at Azbow Pvt Ltd

May. 2024 - Nov. 2024
- Built a Responsive UI: Designed and implemented a dynamic, user-friendly KPI dashboard using Vue.js and Google Charts, enhancing usability and clarity in data visualization.
- API Integration: Integrated RESTful APIs for seamless communication between the front-end and Laravel-based back-end to enable real-time data updates.
- Back-End Development: Developed and maintained a robust back-end infrastructure using Laravel and MySQL, ensuring data integrity and performance scalability.
- Database Design: Designed an efficient database schema with optimized tables to support complex data queries and analytics.
- Data Validation & Debugging: Ensured the accuracy and consistency of KPI metrics through thorough validation techniques, and resolved critical bugs in the application workflow.
- Automation: Implemented and scheduled cron jobs for daily database updates to maintain data freshness and reliability.
- Team Collaboration: Actively participated in client meetings with supervisors and team members to present updates, gather feedback, and align development with client requirements.

Latest Projects

PredictLend

A smart web application that automates the loan approval process using machine learning for credit scoring and risk prediction. It features an interactive dashboard with filters for agent evaluation and customer segmentation, along with dynamic charts for insightful data visualization.

Tech Stack : Vue.js, Google-Charts, MySQL, Laravel, PHP, Postman, TensorFlow, Keras, ML Algorithms

Occluwatch

A multi-camera vehicle tracking system designed to detect road violations with effective occlusion handling using YOLO for detection and ResNet-based ReID for identity preservation. Utilizes Graph Neural Networks and temporal analysis to optimize cross-camera vehicle tracking. Integrated OCR for automated license plate recognition and backend services to issue violation tickets via SMS and email notifications.

Tech Stack : YOLO, ResNet, Vehicle ReID, GNN, OCR, Python, Flask, Spatial temporal analysis, Vue.js

inPath

A web application for students and lecturers to efficiently manage exams. Features include randomized exam hall desk allocation, QR code-based attendance tracking, and machine learning-driven analysis of past results to recommend suitable fields of study.

Tech Stack : HTML, CSS, JS, LogisticRegression, NodeJS, MongoDB Atlas Cluster

Individual Research

Developed a GAN-based image compression system that significantly improves perceptual quality at low bitrates (0.1–0.5 bpp). The model combines an encoder-generator-discriminator setup with saliency maps and multi-component loss functions to enhance important image regions such as faces and text, outperforming traditional JPEG and baseline GAN methods.

Tech Stack : Tensorflow, python, GANs, VGG19, MobileNetV2, Evaluation Metrics: MS-SSIM, LPIPS, FID

Space Sync

This is a mobile application developed using Unity and C#, designed to enhance the furniture shopping experience through Augmented Reality (AR). The app allows users to visualize furniture pieces in their real-world environment before making a purchase, giving them a clear idea of how each item fits and looks in their space

Tech Stack : Unity, c#, Google Cloud

3 Bits

An advanced maze generation and solving application implementing multiple pathfinding algorithms (BFS, DFS, A*, Dijkstra's) with real-time visualization. System integrates computer vision techniques including Canny edge detection, adaptive thresholding, and Hough transform to extract maze structures from images.