Charuka Herath

Charuka Herath is a Doctoral Researcher at the Institute of Digital Technologies, Loughborough University London. With a background in software engineering and cryptography, Charuka brings a blend of practical experience and research innovation to the field of privacy-preserving artificial intelligence. He previously worked as a Senior Software Engineer at Sysco LABS and WSO2 Private Limited and currently contributes to a joint Post-Quantum Cryptography (PQC) research initiative with Airbus Endeavour and Cardiff Metropolitan University. In addition, he lectures in Computer Science and Business Studies at Arden University, London.

Charuka’s research titled ‘Privacy-preserving Machine Learning Techniques in Federated Learning’, focuses on Federated Learning (FL)—a distributed machine learning technique where data remains on users' devices, enhancing privacy and data security. While FL holds transformative potential in privacy-sensitive fields like healthcare and finance, it faces several key challenges in real-world implementation. These include privacy leakage through model updates, inefficiencies caused by heterogeneous devices, and susceptibility to adversarial attacks.

To overcome these, Charuka is developing a novel FL framework that incorporates dynamic client selection and data-entropy-driven scheduling to improve training efficiency and fairness. His research also explores lightweight secure aggregation methods and robust defence mechanisms to guard against malicious participants. These innovations aim to bridge the gap between theoretical advancements and practical deployment of Federated Learning, particularly on mobile, edge, and IoT devices.

Outside his doctoral research, Charuka is deeply committed to educational outreach and community engagement. He serves as the Lead Coding Instructor at Ignite Hubs, a UK-based charity creating learning platforms for children and young people. He also established a STEM after-school club at libraries across Canada Water, Una Marson, and Walworth.

In recognition of his volunteer efforts, Charuka was invited to a garden party at Buckingham Palace on 7 May, by command of His Majesty the King via the Lord Chamberlain.

• Enhancing Federated Learning Convergence With Dynamic Data Queue and Data-Entropy-Driven Participant Selection – IEEE Internet of Things Journal (Featured)
• Recursive Euclidean Distance-based Robust Aggregation Technique for Federated Learning – 2023 IEEE IAS GlobConET
• Correctness Proof for a Ring-LWE Multi-Authority CP-ABE Scheme – IEEE ICDCS 2025
• FheFL: Fully Homomorphic Encryption Friendly Federated Learning with Byzantine Users – IEEE Transactions on AI (Submitted)
• FLSFL: Fully Lightweight Secure Federated Learning with Byzantine Robustness – ICML 2025 CFAgentic Workshop (Submitted)
• QuantFL: Communication-Efficient Federated Learning via Pre-Trained Model Quantization – IEEE MLSP 2025 (Submitted)
• Submitted all published journals to the UK AI Research Symposium