A Scalable DAG based Byzantine Fault-Tolerant Consensus Protocol.

• Mahi-Mahi is a novel randomized BFT consensus protocol that guarantees liveness under asynchronous network conditions.
• Contributed to the design, implementation, and the evaluation of the Mahi-Mahi BFT consensus protocol.
• Mahi-Mahi achieves 20% lower latency than the state-of-the-art asynchronous blockchains.
• Reduced evaluation costs by 20% using an implementation of a novel orchestration framework.
• Github: Mahi-Mahi-Code
• Paper: Mahi-Mahi: Low-Latency Asynchronous BFT DAG-Based Consensus
• Key skills: Distributed Algorithms, BFT, Consensus Protocols, System Design, Software Engineering, Research, Technical Writing, Rust, python, bash, AWS EC2.

A novel testing framework for consensus protocols.

• Consensus Torture is a novel framework to messure the advasarial performance of different consensus protocols.
• Developed a novel, extensible framework for systematically evaluating over 20 different open-source consensus protocols.
• Designed and implemented attack scenarios to simulate network partitions, crashes, and DDoS attacks to evaluate consensus protocols under varying advasarial conditions.
• Key skills: Distributed Algorithms, BFT, Consensus Protocols, System Design, Software Engineering, Performance Measurement, Software Testing, Research, Technical Writing, go-lang, python, bash, AWS EC2.

A novel framework to measure the performance of different blockchain transaction layer architectures.

• This project aims at building a framework to messure the performance of different blockchain execution layers such as Account-based, UTXO-based, and Zero-history-based.
• Designed and implemented a novel performance evaluation framework based on the HotStuff consensus protocol, creating a unified framework for performance analysis across various transaction layer protocols.
• Evaluated six different transaction models, analyzing their latency and throughput performance with respect to payload size, input/output sizes, and arrival rates.
• Key skills: Distributed Algorithms, BFT, Consensus Protocols, System Design, Software Engineering, Performance Measurement, Research, Technical Writing, go-lang, python, bash, cloud.

A novel randomized asynchronous consensus protocol.

• QuePaxa is a novel randomized consensus protocol that guarantees liveness under advasarial network conditoins, while providing high performance under normal network conditions.
• Led the QuePaxa project and contributed to algorithm development, prototype implementation, and performance evaluation.
• Published new findings in ACM SOSP, gaining over 32 stars on Github.
• Github: Quepaxa code.
• Paper: QuePaxa: Escaping the tyranny of timeouts in consensus.
• Key skills: Distributed Algorithms, Consensus Protocols, System Design, Software Engineering, Performance Measurement, Research, Technical Writing, Presentation, go-lang, python, bash, AWS-EC2.

A novel wide-area State Machine Replication protocol.

• SADL-RACS is a novel state machine replication algorithm that provides asynchronous liveness and high scalability.
• Led the SADL-RACS project and contributed to consensus algorithm development, prototype implementation, and the performance evaluation.
• SADL-RACS improves the throughput by a factor of two compared to existing state machine replication protocols, while maintaining robustness under adversarial network conditions.
• Github: SADL-RACS code (previously named Mandator-Sporades).
• Paper: RACS and SADL: Towards Robust SMR in the Wide-Area Network.
• Key skills: Distributed Algorithms, Consensus Protocols, System Design, Software Engineering, Performance Measurement, Research, Technical Writing, go-lang, python, bash, AWS-EC2.

A novel extension of Paxos to provide high robustness.

• Baxos introduces a novel approach to contention by replacing leader election in Multi-Paxos with random exponential backoff, enabling multiple nodes to propose concurrently without relying on a leader replica.
• Led the project and contributed to the development of consensus algorithm, prototype implementation, and performance evaluation.
• Baxos outperforms Multi-Paxos and Raft, achieving 128% higher throughput under delayed view change attacks.
• Paper: Baxos: Backing off for Robust and Efficient Consensus.
• Key skills: Distributed Algorithms, Consensus Protocols, System Design, Software Engineering, Performance Measurement, Research, Technical Writing, go-lang, gRPC, python, bash, AWS-EC2.

• This project implements a novel performance evaluation framework for Java Micro Services.
• Led the design and implementation of the Java-based microservices performance framework.
• Conducted performance analysis, assessing individual microservices as well as different microservices architectural patterns.
• Paper 1: An Analysis of Holistic Tail Latency Behaviors of Java Microservices.
• Paper 2: An Analysis of Throughput and Latency Behaviours Under Microservice Decomposition.
• Github: Python library for statistical analysis of performance data.
• Key skills: Distributed Systems, Cloud, Software Engineering, Performance Measurement, Research, Technical Writing, Java, python, bash, AWS-EC2.

A novel neural machine translation system for Sinhala and Tamil.

• This project proposes a new Nueral Machine Translation system for the low resourced language pair Sinhala and Tamil.
• Led the design, implementation, and evaluation of a novel translation model using the OpenNMT engine.
• Achieved a notable BLEU score despite working with a highly limited corpus size.
• Paper 1: Neural Machine Translation for Sinhala and Tamil Languages.
• Paper 2: Handling Rare Word Problem using Synthetic Training Data for Sinhala and Tamil Neural Machine Translation.
• Paper 3: Transliteration and Byte Pair Encoding to Improve Tamil to Sinhala Neural Machine Translation.
• Key skills: Neural Machine Translation, Research, Technical Writing, Presentation, python.

Please see my google scholar profile