Computer Science & Engineering · University of Moratuwa

Hi, I'm Tharusha Perera.

I'm an undergraduate at the University of Moratuwa specializing in Data Science, working across artificial intelligence, computer vision, and systems engineering. My research looks at how models behave under real-world messiness, from post-crisis time series to cost-sensitive medical diagnosis to multi-source signal fusion, and I build full-stack and embedded systems to put that work into practice.

I'm first author on a MERCon 2026 paper on tourism demand forecasting, and I'm currently looking for research collaborations, internships, and graduate opportunities in AI/ML and data science.

3.67 / 4.00Current CGPA
1stAuthor, MERCon 2026
Dean's ListSemester 1
10+Research & engineering projects
Artificial Intelligence Machine Learning Computer Vision Data Science Computer Architecture

Research

MERCon 2026 · Dept. of Computer Science & Engineering, University of Moratuwa

The Effect of Search Intent Data on Predicting Post-Crisis Daily Tourist Arrivals in Sri Lanka

Tharusha Perera, Janith Mahanama, Maleesha Kumarasinghe, Bimsara Udurawana, Praveen Nawarathna, Patalee Narasinghe, Nathali Athukorala, Sandareka Wickramanayake, Nisansa de Silva

Dept. of Computer Science & Engineering, University of Moratuwa, Sri Lanka

A multi-model forecasting framework for Sri Lanka's post-crisis daily tourist arrivals (2023–2025), integrating official arrivals data with Yandex search intent, localized Google Trends, exchange rates, and weather records across seven model architectures.

8.94%Best MAPE (SVR, RBF kernel)
736RMSE, arrivals / day
27 daysRussian market booking horizon
Read abstract

Accurately predicting daily international tourist arrivals is challenging for tourism-dependent economies like Sri Lanka, especially during its post-crisis recovery (2023–2025). Existing methods often ignore non-Google search engines, missing critical signals from key source markets like Russia, where Yandex dominates. This paper introduces a multi-model forecasting framework that integrates official daily arrivals with Yandex data, localized Google Trends, exchange rates, and weather records. We evaluate seven architectures: Pure SARIMA, SARIMAX, XGBoost, Random Forest, LSTM, Support Vector Regression (SVR), and a Sequential Hybrid, over a 550-day rolling test period. To prevent data leakage, feature selection utilizes Granger causality and cross-correlation analysis strictly on training data. The results demonstrate that SVR with an RBF kernel performs best, achieving a rolling MAPE of 8.94% and an RMSE of 736 arrivals/day. Crucially, Yandex queries for tours and flights emerge as the strongest exogenous predictor across all machine learning models, completely outperforming Russian Google Trends data, which proved uninformative. These findings confirm that aligning predictive features with a tourist market's preferred search engine significantly enhances forecasting accuracy, providing highly actionable insights for modern tourism intelligence systems.

Projects

Research & Machine Learning

Forensic XAI Auditing for Cardiovascular Diagnostics

A cost-sensitive predictive architecture that minimizes clinical false negatives in heart disease triage without distortive synthetic oversampling. Achieved 0.943 sensitivity via F2-score optimization in a 5×5 nested K-fold framework, verified with a 100-seed dual-XAI (LIME + TreeSHAP) forensic auditing protocol.

PythonLIMETreeSHAPScikit-learn
Ongoing

Physics-Informed GNNs for Multi-Horizon Disease Forecasting

A spatio-temporal Graph Neural Network architecture to forecast weekly, district-level dengue incidence across Sri Lanka. Integrates an epidemic-mechanistic (SEIR-SEI) physics-informed loss function, conditional GAN-based data augmentation for scarcity, and a learned adaptive graph to capture unobserved inter-district transmission dynamics.

PyTorch GeometricGCN/GATGANsSpatio-Temporal GNNs
Ongoing

AudioLIT: Multimodal Explainable-AI Workbench for Audio

An interactive, multimodal Explainable-AI (XAI) workbench extending the Learning Interpretability Tool (LIT) for speech recognition (ASR), emotion analytics (SER), and audio deepfake detection (ADD). Features spectrogram attributions, acoustic wave profiling, latent-space embeddings (t-SNE/UMAP), and counterfactual signal mutation.

PyTorchHugging FaceCaptumFastAPIReact

Advanced Spot the Difference Pipeline

A change-detection pipeline for precise localization and object matching in imagery, combining a multi-model ensemble (OWL-ViT + Grounding DINO) with Siamese ViT-based change localization and CLIP-assisted matching. Reached the top 8 at Octwave 2.0.

Computer VisionOWL-ViTCLIP

Infotech RAG Chatbot

A retrieval-augmented chatbot that answers student queries strictly from course materials, cutting LLM hallucination by grounding responses in vector-embedding retrieval.

LangChainChromaDBRAG

Kidney Disease Classification using CT Scans

A CNN-based deep learning classifier that identifies kidney disease from CT imagery to assist early medical diagnosis.

TensorFlowCNNMedical Imaging

Full-Stack & Distributed Systems

Hemas Sync: Enterprise Knowledge Hub

Built at AITHON 2026 (Track 2: D2D Corporate Use Cases) with Team AIvengers, an enterprise workflow that automates SOP generation from spoken knowledge and answers questions with citation-grounded Q&A, via an iterative LangGraph pipeline over Whisper and Gemini. Reached the top 30 teams.

ReactFastAPILangGraphChromaDB

OnTime Transit Platform

A real-time transit tracking and ETA prediction platform for Sri Lankan expressway buses built as a university group project: an 8-service architecture on DigitalOcean Kubernetes with Kafka event streaming and a Kong API Gateway, processing GPS telemetry at sub-2-second latency.

KubernetesKafkaNext.jsPostGIS

High-Performance Order Matching Engine

Built with the LSEG dataset, a distributed client-server matching engine with strict price-time priority, O(log N) insertions, O(1) routing, and zero memory leaks.

C++Boost.AsioDocker

RPAL Interpreter

A university module assignment: a complete C++17 interpreter for RPAL, a functional language used to teach compiler construction, with a lexer, recursive-descent parser, AST standardizer, and a Control Stack Environment (CSE) machine. Achieved 100% behavioral parity with the reference implementation, verified via differential testing.

C++Compiler DesignCSE Machine

IoT & Embedded Systems

Smart Compact Biogas Ecosystem

Built for the SPARK Challenge 2025, a decentralized, AI-powered IoT energy system for compact biogas digesters with sub-500ms ESP32 edge safety loops, LSTM on-device inference, and a real-time Flutter dashboard over MQTT. Advanced to the semifinals as one of the top 8 projects.

FreeRTOSESP32TFLiteFlutter

Nano Processor Design

A reprogrammable 4-bit nano processor implemented in VHDL and deployed on an FPGA, with full digital logic synthesis and datapath design.

VHDLFPGAXilinx Vivado

Experience

Society for Innovation, Engineering and Entrepreneurship (SIEE), Faculty of Engineering, University of Moratuwa Current

Head, Industrial Relations and Education Committee January 2026 – Present

Leading initiatives that connect undergraduates with industry through educational meetings.

Head, Publicity Committee, IESE Grant Scheme Event 2025 January 2025 – December 2025

Directed the publicity campaign for a major grant scheme.

Member, Media and Event Management Committee August 2024 – December 2025

Contributed to the logistical organization of innovation workshops.

Dept. of Computer Science & Engineering, University of Moratuwa Current

Co-Lead & Publicity Strategist, SLIoT 2026 September 2025 – May 2026

Spearheaded the publicity campaign and created video scripts for the department's national-level IoT competition.

Mentoring Events Committee, CSE Student Society (CSESS) October 2025 – July 2026

Volunteering with the student society to bridge the gap between juniors and seniors.

Design Committee, Debug Diaries Podcast

Designed visual assets and promotional materials for the department's official tech podcast.

Volunteer Mentor, Codejam Competition

Mentored students in algorithmic problem-solving during the department's annual coding competition.

IEEE Student Branch, University of Moratuwa Volunteering

Publicity Committee, MoraForesight 3.0 September 2024 – September 2025

Helped organize a residential bootcamp teaching school students about technology and entrepreneurship.

IEEE Computational Intelligence Society (CIS) Current

Head, Finance Committee, IEEE CIS Student Branch Chapter, UoM February 2026 – Present

Directing financial planning, budgeting, and resource management for the university student chapter.

Member, Editorial Committee, IEEE CIS Sri Lanka Chapter February 2026 – Present

Contributing to the section's editorial and content initiatives.

St. Joseph's College, Negombo 2009 – 2022

Head Prefect

Led the student body, ensuring discipline and coordinating major college events.

Secretary, Scout Troop & President, Sinhala Literary Association

Organized community service projects and camp logistics, promoted literary activities and cultural competitions. Also active in Choir, Chess Club, and the Media Unit.

Competitions

October 2025

IEEEXtreme 19.0

Team Codourney · IEEE

Ranked 51st globally out of 19,000+ participants in IEEE's international programming competition.

Education

2024 – 2028 (Expected)

B.Sc. Engineering (Hons) in Computer Science & Engineering

Specializing in Data Science · University of Moratuwa, Sri Lanka

Overall CGPA: 3.67 / 4.00 Sem 1: 3.90 (Dean's List) Sem 2: 3.70 Sem 3: 3.71 Sem 4: 3.46

Relevant coursework: Data Structures & Algorithms, Computer Architecture, Operating Systems, Database Systems, Introduction to Artificial Intelligence, Introduction to Machine Learning, Introduction to Data Science, IoT Devices & Applications, Software Engineering, Programming Languages, Data Communication & Networking, Applied Statistics.

2025 – Present

Chartered Institute of Management Accountants (CIMA)

Management Level · Wisdom Business Academy

Completed Operational Level and Management Level subject exams; awaiting Management Case Study (MCS) exam results.

2023

GCE Advanced Level (Physical Sciences)

St. Joseph's College, Negombo

Z-Score: 2.56 Island Rank: 173 District Rank: 15

Certifications

  • Data Analyst in Python (Career Track)DataCamp, 2025
  • Supervised Machine Learning: Regression and ClassificationDeepLearning.AI, 2025
  • Developing Applications with LangChainDataCamp, 2025
  • Machine Learning FundamentalsDataCamp, 2025
  • AWS Academy Graduate: Cloud Foundations (Training Badge)AWS Academy
  • AWS Academy Graduate: Microservices and CI/CD Pipeline Builder (Training Badge)AWS Academy
  • CIMAOctober 2025
  • The Complete 2022 Web Development Bootcamp
  • Web Design for Beginners (E-Certificate)
  • PHP Tutorial Course

Awards

  • Dean's ListFaculty of Engineering, University of Moratuwa (Semester 1)
  • GCE A/L Island Rank 173, District Rank 15Physical Sciences, National Level

Technical Skills

AI & Machine Learning
TensorFlowKerasXGBoost SVRSARIMAXLIME SHAPTransformers (OWL-ViT) LangGraphLangChainRAG ChromaDB
Software & Web
ReactNext.jsFastAPI Node.jsFlutterPostgreSQL PostGISMongoDBFirebase
Cloud & Architecture
Kubernetes (DOKS)DockerKafka MQTTMicroservicesBoost.Asio
Embedded & Hardware
C++FreeRTOSTFLite ESP32VHDLFPGA Xilinx Vivado
Languages
PythonC++Java JavaScriptTypeScriptDart SQL

Updates

  • August 2026 Published a LaTeX-built academic CV and four job-tailored resumes (ML Engineer, Data Science, Software Engineer, IoT & Embedded Systems), auto-compiled from a single data source.
  • July 2026 Rebuilt this site from the ground up with verified project, research, and experience details.
  • 2026 First-author paper accepted at MERCon 2026: search-intent-driven tourist arrival forecasting for Sri Lanka.

Contact

I'm open to research collaborations, internships, and graduate opportunities in AI, ML, computer vision, and data science. The fastest way to reach me is email.