PhD Candidate · Australian National University

Buddhi Kothalawala

Computer Science · Combinatorial Optimisation · Constraint Programming

I am a PhD candidate in Computer Science at the Australian National University, specialising in branch-and-bound algorithms, combinatorial optimisation, and constraint programming. My doctoral research centres on the maximum common subgraph (MCS) problem — developing novel upper bounds, symmetry-breaking frameworks, and heuristic-exact hybrid methods.

I bring 3+ years of production software engineering experience at WSO2, where I built core components of the Ballerina programming language and the Siddhi stream processing engine.

Combinatorial Optimisation Branch-and-Bound Constraint Programming Symmetry Breaking Reinforcement Learning Graph Algorithms
Download CV
buddhi.kothalawala@anu.edu.au b.wathsala.bw@gmail.com GitHub LinkedIn Scholar Medium
Buddhi Kothalawala, PhD Candidate in Computer Science at the Australian National University
Canberra, Australia PhD Candidate
Australian National University

Education

PhD in Computer Science
2022 — 2026
The Australian National University · Canberra, Australia
  • Thesis: Algorithmic Advances in Exact Maximum Common Subgraph (MCS) Computation
  • Research Focus: Novel branch-and-bound algorithms enhanced with reinforcement learning and symmetry breaking
  • Funding: ANU PhD Scholarship (International) + HDR Fee Remission Merit Scholarship
Bachelor of Science in Computer Science (Honours)
2015 — 2019
University of Colombo School of Computing · Colombo, Sri Lanka
  • Class: First Class Honours
  • Thesis: Online Learning for Solving Data Availability Problem in Natural Language Processing
  • ICTA Award for Best Computer Science Final Year Research Project (2018)

Research & Publications

Accelerating Maximum Common Subgraph Computation by Exploiting Symmetries

Kothalawala, B., Koehler, H., & Farhan, M.
ACM SIGMOD 2026 · CORE A* Published
  • Developed a symmetry-breaking framework for MCS algorithms substantially outperforming state-of-the-art methods
  • Demonstrated significant speedups on standard MCS benchmarks through comprehensive symmetry-aware pruning
PDF Code

From Exploratory Heuristics to Exact Search: Accelerating Maximum Common Subgraph Algorithms

Kothalawala, B., Koehler, H., Wang, Q., & Farhan, M.
Australasian Database Conference (ADC) 2025 Published
  • Proposed a novel heuristic mechanism to escape local optima in depth-first branch-and-bound algorithms
  • Developed a principled transitioning criterion from heuristic to exact search with theoretical convergence guarantees
PDF Code

Learning to Bound for Maximum Common Subgraph Algorithms

Kothalawala, B., Koehler, H., & Wang, Q.
CP 2025 — 31st International Conference on Principles and Practice of Constraint Programming · CORE A Published
  • Developed a reinforcement learning framework achieving maximum possible bound reduction for NP-hard optimisation problems
  • Empirically demonstrated superior performance over existing branch-and-bound approaches
PDF Code

Online Learning for Solving Data Availability Problem in Natural Language Processing

Kothalawala, B., Weerasinghe, R., & Kumarasinghe, P.
NL4AI @ AI*IA 2019 Published
  • Addressed model retraining challenges for the low-resource language Sinhalese
  • Applied online learning with CRFs, RNNs, and bidirectional LSTMs to overcome data scarcity in NLP
PDF Code

Experience

Research Assistant (Part Time)
Jun 2026 — Oct 2026
UNSW Canberra, Australian Defence Force Academy · Canberra, Australia
  • Conducted research on graph theory and subgraph-based algorithms applied to supply chain networks
  • Performed theoretical analysis and experimental evaluation of these algorithms in C++ and Python
Academic Tutor
Feb 2023 — Present
The Australian National University · Canberra, Australia
  • Taught and mentored 150+ students across three advanced courses with consistently high satisfaction ratings
  • Courses: Statistical Machine Learning (COMP4670/8600), Relational Databases (COMP2400/6240), Data Wrangling (COMP3430/8430), TechLauncher (COMP3500/COMP4500/COMP8715)
PhD Candidate & Researcher
Jun 2022 — June 2026
The Australian National University · Canberra, Australia
  • Pioneering exact solutions for NP-hard combinatorial optimisation using advanced constraint programming
  • Developing novel branch-and-bound algorithms enhanced with symmetry breaking and reinforcement learning
  • Published at top-tier venues: SIGMOD 2026, CP 2025, and ADC 2025
Senior Software Engineer
Nov 2020 — Jun 2022
WSO2 · Colombo, Sri Lanka
  • Owned development of Ballerina's standard libraries: gRPC, Time, and I/O — driving critical feature implementation
  • Authored technical articles to enhance community awareness and adoption
Software Engineer
Feb 2019 — Oct 2020
WSO2 · Colombo, Sri Lanka
  • Developed the Siddhi Kubernetes operator and cloud-native features for the Siddhi stream processing engine
  • Led the migration of Ballerina from a monolithic repository to a scalable multi-repository architecture
  • Established CI/CD pipelines and best practices across multiple engineering teams
Software Engineer Intern
Aug 2017 — Jan 2018
WSO2 · Colombo, Sri Lanka
  • Developed an internal application for license and repository management
  • Gained hands-on experience across the full software development lifecycle

Selected Projects

Ballerina Multi-Repository Architecture Migration

  • Led the transition of Ballerina from a monolithic codebase to a multi-repository architecture to improve maintainability and scalability
  • Coordinated with multiple teams to guide and support migration of their respective components
  • Established CI/CD pipelines and best practices adopted across the organisation

High-Performance Standard Libraries — Ballerina & Siddhi

  • gRPC Library: Implemented key features including message compression for Ballerina's gRPC standard library
  • I/O Library: Simplified I/O APIs in Ballerina to improve scalability and developer experience
  • Time Library: Developed UTC time handling and timezone conversion support
  • Siddhi Kubernetes Operator: Made Siddhi cloud-native by developing Kubernetes and Docker deployment features

MCS Algorithm Research Suite

  • Novel bound calculation method achieving maximum empirical bound reduction in branch-and-bound MCS search
  • Novel heuristic to enhance lower bound quality, enabling efficient transition to exact search
  • Comprehensive symmetry-breaking framework to reduce redundant search effort in MCS algorithms

Awards & Recognition

Community & Service

Open Source
Core Contributor — Ballerina programming language, Siddhi streaming ecosystem, and StackOverflow (1K+ reputation)
Community Leadership — Conducted the first Ballerina Language Community Call, reaching 1,200+ views
Technical Writing — Published articles on Medium and DZone with 20,000+ total views
Mentoring & Teaching
Research Supervision — Mentored a final year research project on compressing Bi-LSTM-based contextualised word embeddings
Industry Mentorship — Guided junior developers at WSO2 on best practices and career development
Academic Tutoring — Delivered tutoring to 150+ students across three university courses at ANU
Professional Development
Toastmasters International — Completed three levels of the Dynamic Leadership Pathway Programme
Conference Presentations — Speaker at technical conferences and community events
Media Engagement — Featured expert on the Internet Sampath Android OS Radio Programme