Pasindu Nivanthaka Tennage
pasindu.tennage@epfl.ch +41 21 693 30 95 https://pasindutennage.github.io/
Citizenship: Sri Lanka
Fields of expertise
Consensus, Blockchains, Distributed Algorithms, BFT, Scalability, Software Engineering, System Design, Software Testing, Performance Optimization.
Biography
I am a distributed systems engineer and researcher with a strong background in designing and evaluating scalable, resilient systems.My journey began at the University of Moratuwa, Sri Lanka, where I earned my degree in Computer Science and conducted early research on neural machine translation for low-resourced languages. I continued at the same university for my Master’s degree, focusing on distributed and cloud computing. My Master’s work included load balancing, micro-services performance characterization, and server architecture design.
In 2019, I began my PhD at EPFL, Lausanne, Switzerland, specializing in robust and resilient consensus protocols. My research involves designing new consensus protocols, implementing them, and performing real-world evaluations.
Education
Master’s Degree, Computer Science and Engineering
Specialization in cloud computing, GPA 4.2/4.2
University of Moratuwa
2019
Bachelor’s Degree, Computer Science and Engineering
GPA 4.13 / 4.2
University of Moratuwa
2017
Awards
2018 : Gold medal for the highest GPA : University of Moratuwa
Publications
Selected publications
| Pasindu Tennage, Cristina Basescu, Lefteris Kokoris-Kogias, Ewa Syta, Philipp Jovanovic, Vero Estrada-Galinanes, Bryan Ford SOSP 2023 |
QuePaxa: Escaping the tyranny of timeouts in consensus |
| Pasindu Tennage, Srinath Perera, Malith Jayasinghe, Sanath Jayasena, HPCC 2019 |
An Analysis of Holistic Tail Latency Behaviors of Java Microservices |
Research
Consensus-Torture
I developed a novel and extensible framework for systematically evaluating over 20 open-source distributed consensus protocols. As part of this work, I designed and implemented attacks to simulate network partitions, crashes, and DDoS scenarios, enabling comprehensive evaluations of protocol performance and resilience.Mahi-Mahi: A scalable DAG-based consensus protocol
I contributed to the design, implementation, and evaluation of a secure and scalable distributed consensus protocol using Rust. This protocol, Mahi-Mahi, achieves 20% lower latency compared to state-of-the-art asynchronous blockchains. Additionally, I reduced evaluation costs by 20% by implementing a workflow orchestration system to streamline the evaluation process.Paper: Mahi-Mahi: Low-Latency Asynchronous BFT DAG-Based Consensus
Code: https://github.com/PasinduTennage/mahi-mahi-consensus
SADL-RACS state machine replication
I led the SADL-RACS project, where I contributed to the design of consensus algorithms, the implementation of prototypes using Go, and the performance evaluation of the system on AWS EC2. SADL-RACS improves the throughput by a factor of two compared to existing state machine replication protocols, while maintaining robustness even under adversarial network conditions.Paper: RACS and SADL: Towards Robust SMR in the Wide-Area Network
Code: https://github.com/ISTA-SPiDerS/Mandator-Sporades