Janith Chathuranga Mahanama
Computer Science & Engineering, University of Moratuwa
About
3rd-year Computer Science and Engineering undergraduate at the University of Moratuwa, specializing in Data Science and Engineering. Focused on forecasting, explainable AI, and full-stack systems, with research accepted at MERCon 2026 and hands-on work spanning machine learning, infrastructure, and product engineering.
Education
University of Moratuwa
2024 – PresentB.Sc Engineering (Hons), Computer Science and Engineering
Moratuwa, Sri Lanka- CGPA: 3.95 / 4.00
- Stream: Data Science and Engineering
- Dean's List, SGPA Sem 1: 4.00 · Sem 2: 4.00 · Sem 3: 3.92 · Sem 4: 3.91
Trinity College Kandy
2009 – 2023Primary and Secondary Education
Kandy, Sri Lanka- GCE Advanced Level (Physical Science) 2022/23, Z-Score 2.6570 · District Rank 10 · Island Rank 106
- GCE Ordinary Level 2019, 8 A's, 1 B
Research Experience
Predicting Post-Crisis Daily Tourist Arrivals
Accepted, MERCon 2026- Researched "The Effect of Search Intent Data on Predicting Post-Crisis Daily Tourist Arrivals in Sri Lanka," building a multi-model forecasting framework (SARIMA, SARIMAX, XGBoost, Random Forest, LSTM, SVR, Sequential Hybrid) integrating official arrivals with Yandex search queries, Google Trends, exchange rates, and weather data.
- Benchmarked forecasting architectures and conducted feature engineering and explainability analysis, achieving a rolling MAPE of 8.94% with an RBF-SVM model.
Feature Selection Research (NIPS 2003)
Not Published- Compared filter, wrapper, embedded, and hybrid feature selection methods on the high-dimensional Madelon dataset from the NIPS 2003 Feature Selection Challenge.
- Implemented and benchmarked selection pipelines that improved balanced accuracy from 0.58 (baseline, all features) to 0.89 using a hybrid approach.
Projects
AudioLens, Explainable AI Research Workbench for Speech Models
Ongoing- Tech: Next.js, FastAPI, Celery, Redis, Docker, PyTorch, Hugging Face Transformers.
- A research workbench for explainable AI in speech recognition (Whisper, Wav2Vec2): saliency and attention-based attribution, perturbation-driven robustness testing, layer-wise representation probing, internal activation and Jacobian-based sensitivity analysis, Whisper hallucination detection, and linguistic-vs-acoustic influence and accent/language fairness analysis, all served through an asynchronous, containerized Celery/Redis job architecture.
- Researchers can upload custom speech models and datasets to inspect the blackbox end-to-end: tracing a bad prediction back to the exact audio evidence or acoustic condition that caused it.
Smart Waste Management System
2026- Tech: GitOps via ArgoCD, Kafka/EMQX event streaming, Kong API gateway, Keycloak/Vault identity & secrets, Prometheus/Grafana monitoring, Hyperledger Fabric ledger for tamper-proof records.
- An IoT-integrated platform to monitor real-time waste levels and optimize collection routes across edge, IoT, mobile, and analytics components.
- Worked in the DevOps and Platform Infrastructure team.
Pharma Availability Control Tower
2026- Tech: FastAPI, Next.js, SQLAlchemy, MySQL, XGBoost, OR-Tools, LangGraph, LangSmith, Chroma, Gemini, Sentence-Transformers.
- A pharmaceutical distribution planning platform consolidating shortage forecasting, shipment prioritization, dispatch planning, and execution tracking into a single planner workflow.
- Built the ML module: priority-scoring for inbound manifest clearance, an XGBoost 48-hour shortage forecasting model, and an OR-Tools dispatch optimizer, plus a multi-agent RAG chatbot (LangGraph supervisor routing to knowledge, operations, and analytics agents over Chroma) for conversational querying of policies, stock, forecasts, and dispatch plans.
RPAL Interpreter
2026- Tech: Java.
- Built a complete interpreter for RPAL, a functional programming language: lexing, recursive-descent parsing, AST standardization, and evaluation via a Control/Stack/Environment (CSE) machine, without external lexer/parser generators, supporting closures, recursion, tuples, and built-in operators.
Banking Management System
2025- Tech: Next.js, Express.js, Neon Postgres, Docker.
- A digital banking platform for a microfinance institution: savings accounts, fixed deposits, automated interest accrual, and real-time transaction processing across multiple account tiers.
- Contributed to full-stack development and a CI/CD workflow automating build verification and unit testing.
12-bit Nanoprocessor
2025- Tech: VHDL, Xilinx Vivado, FPGA.
- Designed a custom 12-bit nanoprocessor: ALU, register bank, instruction decoder, verifying functionality through simulation and hardware synthesis.
Competitions
AITHON 2026 by Hemas Holdings PLC
2026- Top 10 Finalist.
IEEEXtreme 19.0 Programming Competition
2025- Global Rank 51.
Coderally 6.0, Hackathon by IEEE Computer Society of IIT
2025- 4th Place (Advanced Tier).
CodeArena, Coding Competition by IET on Campus KDU
2025- 4th Place (Finalist).
Devthon 2.0, Web App Development Competition
2025- Semi-Finalist with "Safe Steps: Women Safety Website."
Certifications
- AWS Academy Graduate, Cloud Foundations
- AWS Academy Graduate, Microservices and CI/CD Pipeline Builder
- Machine Learning Specialization, Coursera (Ongoing)
- Python Programming Online Course, University of Moratuwa
- Web Development Online Course, University of Moratuwa
- Level 3 Certificate in Information Technology, School of Computing, Esoft Metro Campus
Experience & Extracurriculars
University of Moratuwa
2024 – Present- Member of IEEE Student Branch and IESL.
- Member, Strategic Events Committee, Computer Science & Engineering Student Society.
- Member, Publicity Committee for SLIoT 2026.
Trinity College Kandy
2009 – 2023- Teaching Assistant, Advanced Level Section (2023).
- Committee Member, Social Service Union (2019), Science Society (2015, 2017).
- Member, Astronomical Society, Automotive Club, Design Club.
- School Hockey Team (2013–2014, 2016–2017).