Hello world!
I'm Tharuka Pavith,
a Software Engineer.

Know more

About me

Profile Image

I'm Tharuka Pavith, a final year undergraduate with a strong background in software engineering and machine learning. I'm currently studying at the Department of Electrical and Information Engineering, University of Ruhuna. My enthusiasm for Software Engineering and fields of Artificial Intelligence drives me to develop innovative solutions.

Experience

Codegen International (Pvt) Ltd.

(from October, 2023 to January, 2024)

I gained my first training experience as a software engineer at CodeGen International. During my time at the company, I acquired knowledge and experience in Agile methodology (Scrum), Java Spring Boot, Angular framework, Machine Learning (Intent classification), and Chatbot development using LangChain.

Projects

GAN-Based Text-to-Image Synthesizer

Developed a Generative Adversarial Network (GAN) for text-to-image synthesis, integrating it with an LLM-based chatbot and a web application. From this project I was able to improve my knowledge in deep learning and development of AI chatbots using LangChain.

Solar Power Generation Prediction for a Power Plant

Developed a machine learning model to predict solar power generation using regression techniques. I was able to enhance my skills in data pre-processing, exploratory data analysis, and model evaluation. Specially focused on optimizing predictions by tuning hyperparameters and feature selection.

Binary Classification of Poisonous Mushrooms

Implemented several classification models like RandomForest, KNN, XGBoost and Deep Neural Networks for binary classification of mushrooms, focusing on feature encoding and model evaluation using the Matthews Correlation Coefficient (MCC). This reinforced my knowledge in data pre-processing, exploratory data analysis and model assessment.

Contact

Thank you for visiting my portfolio! I'm currently setting up the best ways for you to get in touch with me. Please check back soon for updates, or connect with me on LinkedIn in the meantime.