SelfUp
Technologies - Javascript
SelfUp is a Chromium based extension that I developed entirely from scratch using vanilla
JavaScript. It delivers daily motivational quotes and offers a variety of royalty-free
music tracks - such as concentration, meditation, and cheerful tunes - for users to
listen to while working. This simple extension eliminates the need to search through
streaming sites for music, by reducing distractions while on focus. This project
enriched my skills in end-to-end extension development.
Invity
Technologies - React, Node.js, Firebase
Invity is an automated bulk invitation generator. I started this as a side project and
developed the completed platform in less than a month. Through this project, I learned
to create a user-friendly platform where people can easily upload an invitation image,
personalise it using built-in tools, and, here's the cool part, generate a bunch of
customised invitations with just one click. It was an exciting journey, and I gained
valuable skills in simplifying processes and making tasks more efficient. I believe this
project showcases my ability to create practical solutions that bring real value to
users.
Hybrid-Attention End-to-End Speaker Diarization
Final Year Project
Technologies - Deep Learning, PyTorch, Python, GCP
This was based on implementation of the state-of-the-art model for 2 speaker overlapping
speech diarization by reducing the Diarization Error Rate (DER). We proposed three more
models by improving the current state-of-the-art Self Attention
model, which are combinations of Local Dense Synthesizer-Attention layer with Self
Attention, Relative Attention and External Attention layers.
Search Engine Politicians Sinhala
Technologies - Python, Flask, React.js, BeautifulSoup, Elasticsearch, Kibana
This is an information retrieval system that can be used to query and get any type of
data about Sinhala Politicians. The dataset of the project was scrapped using
BeautifulSoup and the project was implemented using Python, React.js, and supported by a
python-Flask powered backend. I have used Elasticsearch for indexing and quering.
DrivenData | Pump-It-Up | ML Project
Technologies - Python, Jupyter Notebook
Python solution for Pump-It-Up, DrivenData machine learning competition. I have used
various kinds of encoding, normalization techniques and learned pre-processing and
post-processing techniques as well. I was within 4% of the global participant results in
this competition. The result was great because it had been 0.8234, compared with the
first place 0.8294.
Distributed Chat System
Technologies - Java
This group project is about creating a distributed chat application. The system consists
of two main distributed components: chat servers and chat clients, which can run on
different hosts. I have implemented all the functions for one server
(createroom,joinroom,etc.) and implemented the Heartbeat algorithm using gossiping and
consensus for failure detection.
Kritify | Mobile Application
Technologies - Android Studio, Java, Spring Boot
This is a social media mobile app that hosts all your reviews about business entities in
one place. This is an Android app integrated with HMS core. I have implemented the
functionalities for Huawei Account Kit, post creation, and point reviewing for each
visited place.
TravelME
Technologies - Flutter, Dart
Implemented an "Online tour planning platform" that can be used to plan a tour. The user
can find the most attractive places, nearest places and hotels around the city as well.
The mobile application frontend was implemented by me using Flutter framework.