machine learning for systems
I build learning-based schedulers that make cloud and distributed systems faster and more energy-efficient.
My MSc research built graph-neural-network and deep-reinforcement-learning schedulers for cloud workflows that balance energy use, completion time and quality of service. On real Google cluster traces, the latest one outperforms heuristic and evolutionary baselines in hypervolume and IGD. Before that, at WSO2, I built the REPL that now ships with the Ballerina compiler. I am also co-founder and CTO of iXD Labs, a software company in Sri Lanka. I am looking for funded PhD positions.
- publications
- 4 journal papers, incl. IEEE Access and CS Review
- compilers
- Built Ballerina Shell, shipped with the Ballerina compiler
- education
- MSc by Research; BSc First Class, GPA 4.08/4.20
- industry
- Co-founder & CTO, iXD Labs, since 2022
News
- Survey “Energy-Efficient Multi-Objective Cloud Workflow Scheduling: Taxonomy and Techniques” published in Computer Science Review, vol. 63.
- Survey “Energy-Efficient Multi-Objective Cloud Workflow Scheduling: Taxonomy and Techniques” accepted for publication in Computer Science Review.
- Graduated with an MSc by Research from the University of Moratuwa.
- Two-stage deep RL + GNN workflow scheduler published in IEEE Access.
- Paper on energy-efficient dynamic workflow scheduling published in Sensors.
Selected publications
all publications →- Computer Science Review 63 · 2027Energy-Efficient Multi-Objective Cloud Workflow Scheduling: Taxonomy and Techniques
- IEEE Access 14 · 2026Multi-Objective Cloud Workflow Scheduling with Two-Stage Deep Reinforcement Learning and Graph Neural Networks
- Sensors 25(5) · 2025Energy-Efficient Dynamic Workflow Scheduling in Cloud Environments Using Deep Learning