Buddhi Kothalawala

I use past experiences to do present experiments in order to make a better future.

Email: b.wathsala.bw@gmail.com
Phone: +94776652085
Course: CS

Interests: R&D, Machine Learning, Operating Systems, Compilers and Automata Theory, Quantum Computing, Natural Language Processing, Networking

Technical Skills: Python, Java, PHP, C/C++, Ballerina, Go, Linux, React JS, Material UI, Git, WSO2 BPS/MSF4J, REST APIs (Jenkins. Nexus and GitHub), MySQL

Project Experience

License and Repository Management Application and Maven Plugin
Developed an internal application for WSO2 considering 3 main functionalities – Repository
creation, Library approval, and License generation

WEBMIS – Web-based application for a Career Training Center
Web application for manage internal works and an Android mobile application for students

Braille Reader – Imagine Cup Final Selected Project (2015)
Software for convert text to speech

Middleware using Go language
Develop simple message oriented middleware using Go language

Online Event Bash
People can publish events and also they can gather information about events

Work Experience

  • WSO2 Lanka (Pvt) Ltd – Software Engineer Intern (August 2017 – January 2018)

Achievements | Awards

  • Imagine cup finalist in 2015
  •  Nominated for best Intern at UCSC – 2017
  • IEEE Xtream 11.0 in 2017 – Island rank 47
  • Represented to a cadet platoon for a President guard in 2008

Final Year Project

Online Learning Algorithms for Solving Data Availability Problem in Natural Language Processing

Machine learning models need large data corpus to obtain better accuracies. In practical scenarios this large data corpus available as a stream with respect to the time dimension instead of a whole batch. Most NLP problems suffer from this problem. Thus we apply online learning base machine learning algorithms to solve this data availability problem. The solution of this research acquires from several machine learning models such as online learning, imitation learning, reinforcement learning, and deep learning. Our solution strategy based
on with the bidirectional LSTM architecture.