Govarthenan Rajadurai
Data Scientist & AI Engineer
Co-Founder & Director at DreamSpace Technologies. Building AI systems for social impact, specializing in NLP for low-resource languages and responsible AI. Leading hate speech detection initiatives and LLM bias research in Sri Lanka.
Current Work
DreamSpace Technologies
Co-Founder & Director, Data Scientist
January 2025 – Present
Co-founded and leading a software development agency pioneering AI-driven solutions for social impact. First company to host a hackathon in Batticaloa. Partnering directly with USAID and UN on technology projects while developing local talent through industry-standard practices.
Digital Hate Speech Detection & HateLens.lk
Leading Sri Lanka's most comprehensive digital hate speech detection initiative.
- Constructed Sri Lanka's largest digital hate speech dataset: 50,000+ entries across Sinhala, Tamil, and Latin script with 12 fields
- Managed 20+ person data collection teams
- Fine-tuned two open-weight LLMs for hate speech detection and counter-narrative generation
- Building HateLens.lk API to monitor and analyze hate speech across social media platforms with automated reporting
- Funded by USAID, acknowledged by UN, UNDP, and Global Communities
Tech: RoBERTa, transformers, PyTorch, axolotl, DVC, DagsHub
LLM Bias in Controversial Socio-Historical Topics
Leading white paper investigating bias in popular chatbots and LLMs when handling controversial topics related to Sri Lanka's troubled past. Extensive research using various prompting techniques, user roles, and conversation languages.
Rupeeka: Economic Fact-Checking Chatbot
Developing an intelligent chatbot that verifies economic claims against official Sri Lankan government sources (Central Bank, Ministry of Finance).
- Real-time policy verification across English, Sinhala, and Tamil
- Translates complex economic information into accessible language
- Combats economic misinformation with source-backed responses
Tech: RAG with query expansion, relevance tuning, reranking, langchain
Experience
DreamSpace Academy
Data Scientist (Part-Time)
April 2022 – December 2024
Non-profit social enterprise tackling socio-economic and environmental challenges in Sri Lanka.
- Architected next iteration of AI hate speech detection and secured funding
- Led first AI-powered native hate speech detector in collaboration with Omdena (first tool in Sri Lanka to use language models)
- Developed ML models spanning computer vision, NLP, time series, and recommender systems
- Designed software development infrastructure for DreamSpace Academy's software lab
- Conducted AI and NLP workshops, trained and supervised interns
UCSC Sustainable Computing Research Group
Web Scraping Engineer (Part-Time)
November 2022 – April 2023
- Developed advanced web scraping pipelines for digital forensics research
- Extracted human-readable text from structured and unstructured data
- Preprocessed and parsed data for ElasticSearch analysis
Elimo IT
Automation Engineer
March 2023 – August 2023
- Engineered Excel automation solution for HSBC Bank using Python and win32 API
- Implemented NumPy vectorization for complex formulae optimization
- Designed two-way look-up table to simplify KPI calculation processes
Agrivero.ai
Data Science Intern
September 2021 – January 2022 | Berlin, Germany
AI-powered green coffee grading using computer vision.
- Developed ML model to classify coffee bean sizes based on sub-centimeter differences
- Identified hardware issues through diagnostic analysis, enabling company-wide hardware overhaul that significantly boosted accuracy
Freelancing
Data Scientist & ML Engineer
June 2022 – Present
- Customer retention analysis for Australian supermarket chain
- Financial ticker prediction using time series and LSTM models (German client)
- Web scraping and NLP pipeline for Australian IT support agency
Technical Skills
Languages & Core Tools
Python
SQL
Linux/Unix
Git
DVC
ML & Deep Learning
PyTorch
scikit-learn
scipy
axolotl
NLP & LLMs
transformers
langchain
langsmith
NLTK
Data Engineering
NumPy
Pandas
Polars
OpenCV
Visualization & Web
Matplotlib
Plotly
Streamlit
FastAPI
Flask
Cloud & Infrastructure
AWS
Azure
Docker
Education
Bachelor of Science in Information Systems
University of Colombo School of Computing
Graduated June 2025 | GPA: 3.1/4.0
Recognition & Community
Panel Speaker - APPRAC 2024, University of Vavuniya
Participant - Build Peace 2024, Manila, Philippines
Member - Community of Practice, UNDP Sri Lanka (since ~2023)
Responsible AI Fellowship - Stimson Organization
AWS Cloud Technical Essentials - Amazon Web Services