Skip to content
Ravindu Weerasinghe

About Me

I am Ravindu Tharuka Weerasinghe, an Artificial Intelligence undergraduate at the University of Moratuwa, with a strong focus on building practical and efficient AI systems.

My work spans small language models, MLOps, and real-world system design. I am particularly interested in how AI systems move beyond isolated models into real-world environments, especially in robotics and edge systems.


Experience

WSO2 - Research & AI Intern

  • Developed and fine-tuned small language models using techniques such as LoRA, adapters, and prefix tuning
  • Evaluated models using RAGAS metrics, improving performance of a local model (~60% - ~65%)
  • Built automated MLOps pipelines for training, tracking, and deployment using MLflow, RunPod, and HuggingFace
  • Focused on privacy-sensitive and resource-constrained AI applications

Education

University of Moratuwa

B.Sc. (Hons) in Artificial Intelligence

CGPA: 3.86 / 4.00

Dean's List: 4/6 semesters

* Strong foundation in Neural Networks, Machine Learning, NLP, and Software Engineering.

BCS, The Chartered Institute for IT

BCS Level 6 Professional Graduate Diploma in IT

* Gained knowledge in Advanced Database Management Systems, Big Data Management, Programming Paradigms, and Network Information Systems.


Research

Type II Diabetes Risk Prediction using Multifactor Machine Learning

IEEE Conference Publication (2026)

  • Developed a LightGBM-based prediction system using ~100,000 NHANES records integrating laboratory and lifestyle features
  • Designed a multi-stage feature selection pipeline reducing ~5000 features to 43 clinically relevant predictors
  • Achieved 95.27% ROC-AUC with 88.48% sensitivity and 87.19% specificity on real-world test distribution
  • Built a practical system capable of handling missing data without imputation, enabling real-world deployment

Small Language Models for Privacy-Critical Applications

Independent Review (2025)

  • Explored compression techniques such as quantization, distillation, and pruning
  • Analyzed privacy-preserving techniques (DP, SMPC, FL, HE) for SLMs
  • Investigated future trends including agentic and multi-agent SLM systems

Interests


Personal Note

I enjoy building systems that go beyond theory and work in real-world settings. One of my most impactful experiences was developing a fully autonomous chess-playing machine, which sparked my interest in robotics and hardware-integrated AI systems.

Outside of work, I enjoy badminton, playing guitar, and reading mystery novels.

Feel free to explore my projects or reach out if you'd like to connect.