🛺Hello! I am a Research Scientist at Accenture and a PhD graduate from the Lab for Population Intelligence at RIT, led by Professor Christopher Homan. My research focuses on ethical AI and the prediction of human disagreements in annotation using machine learning. Rather than treating annotation noise as a problem, my work leverages human rater disagreements to build more socially representative and bias-aware datasets. This is especially important, as recent advances in machine learning have uncovered bias risks for marginalized groups.
During my PhD, I’ve interned at Amazon Ads (2023), Meta (2022), and RPI (IBM Watson Project, 2019), where I developed models for handling weak human judgments and well-being in creator communities.
Beyond my research at RIT, I actively work with the University of Kelaniya in Sri Lanka to develop a national electronic medical record system, enabling scalable healthcare technology in resource-limited settings. I also have a background in sociolinguistics, with work on the evolution of Sri Lankan English across generations.
I am passionate about DevOps, building robust, end-to-end research systems. When not at my desk, I love cycling, photography, and discovering new places on my travels.
PhD in Computer Science
Rochester Institute of Technology
BSc in Computer Science
University of Kelaniya
Using LLMs to identify high-quality human annotators by checking if their labels are consistent with AI predictions—helping build better training data while preserving diverse …
Mixing outputs from multiple LLMs (GPT-4, DeepSeek, etc.) using genetic algorithms to evolve better solutions for complex reasoning tasks—like breeding the best answer from diverse …
ProRefine automatically improves AI prompts during inference by having one AI agent give feedback to refine another agent's prompts—boosting accuracy by 3-37% and helping smaller …
We ran a massive experiment: 9 different AI content moderation systems analyzed 92 million YouTube comments about US politics. The results were shocking—different AI systems …