Building interpretable deep learning algorithms for culture-free early detection of antibiotic drug resistant bacteria using Computer Vision and Quantitative Phase Microscopy (QPM). Microscopic images of different bacteria strains are used to train Capsule-Network models with a smooth latent space representation.

A data access framework to simplify the development of real-time bio-signal analysis workflows. It provides (1) a metadata schema to describe datasets and streams, (2) language-bindings to parse metadata, (3) a web app to package datasets with metadata, and (4) a unified API to stream data from bio-sensors, datasets, and simulations alike. Streaminghub was succesfully used to develop eye movement analysis workflows and EEG analysis workflows.

A web service to generate robust links for web resources. It allows to upload PDF documents, extract URLs from PDF annotations/text, invoke the Robust Links Service on URLs to preserve their content in web archives, and thereby generate robust links. Upon generating robust links, their availability can be announced to external systems via Linked Data Notifications. It also allows to view PDF documents side-by-side with robust links.

Collaborative project for webcam-based, multi-user eye tracking on the browser. It eliminates the need for additional eye-tracking equipment by performing facial detection, facial landmarks detection, iris detection, and gaze estimation through webcam feeds.

Using EEG and Thermography data to predict a diagnosis of ASD. Here, EEG signals were first pre-processed. Simultaneously, features were extracted from Thermography data. Next, a wavelet transformation was applied on each EEG signal to generate features. Machine learning models were then trained from these features to predict a diagnosis of ASD.

Estimating the rate of change of the academic web using archived copies of web pages. Here, archived copies are first obtained from the Internet Archive through its CDX Server API. Next, for each web page, archived copies are analyzed to detect when changes have occurred. These changes are then modeled as a poisson process, and used to predict when a page is likely to change in the future.

Built a mobile application for making doctor consultations. Re-structured cloud infrastructure on AWS to reduce costs. Set up unified logging using AWS CloudWatch. Created Dockerfiles and BitBucket Pipelines to build, test, and push build artifacts into AWS. Provided development and production support for the web apps of The GP Service (UK) Ltd.

• Added single-sign-on capability to web portal.
• Upgraded web portal from Angular 2.x to 4.x.

Created Dockerfiles for each core component in Apache OODT. Added a Maven Archetype to auto-generate Dockerfiles and a Docker-Compose file when creating a new Apache OODT project. Issue tracking was handled using JIRA, and documentation was written using Confluence.

A full-stack, event-driven, microservices framework for Java, with business modeling. It abstracts away the complexities of microservices architecture by providing interfaces to write business logic. The framework internally handles microservice lifecycle, data partitioning, and auto-scaling. It also facilitates generating boilerplate code from Business Process Models.

Internship project at WSO2 Lanka (PVT) Ltd to automate question answering in WSO2 Support portal. Created a document index using ElasticSearch, a reinforcement learning model using TensorFlow to predict document-query relevance, and a search interface using HTML, JS, and CSS.

A competition hosted by DrivenData, to model the spread of Dengue using weather and geographical information. Built a machine learning model that ranked #1 in 2017.

A web platform aimed at tourists and tour providers. Tourists can discover tour services and make reservations, while tour providers can post and advertise their services.

A real-time customer satisfaction monitoring framework for the hospitality industry. It uses camera feeds to detect customer faces, analyze their emotions, and provide satisfaction ratings and dissatisfaction alerts in real-time.

A Point-of-Sales, Inventory, Reporting, Loan, and Payroll System for Dambulla Dedicated Economic Centre, Sri Lanka

A web platform to manage information of a medium sized music school. It supports classroom management, progress tracking, attendance, and payroll.