Dakai Kang

Dakai Kang

Research Scientist at Sei Labs. Byzantine Fault Tolerant consensus and distributed systems.

Publications

You can also find my articles on my Google Scholar profile.

Conference & Journal Papers

Cassandra: Consensus with Partial Progress via Robust Partitionable View Synchronization

Published in Proceedings of the VLDB Endowment (VLDB 2027) — to appear, 2027

A consensus protocol that keeps partial progress during network partitions via a two-tier certification framework and a partition-tolerant pacemaker. Accepted at VLDB 2027.

Recommended citation: Xie, S., Kang, D., Chen, J., Gupta, S., Hughes, D. P., & Sadoghi, M. (2027). "Cassandra: Consensus with Partial Progress via Robust Partitionable View Synchronization." Proceedings of the VLDB Endowment (VLDB 2027). (To appear.) arXiv:2607.02856
Download Paper

FEAT: Fair and Efficient Adversarial Transaction Ordering

Published in Proceedings of the ACM Symposium on Principles of Distributed Computing (PODC '26), pp. 489–499, 2026

An AVL-tree-based fair-ordering algorithm that achieves the optimal Θ(|T| log |T|) comparator-query complexity for γ-Batch-Order-Fairness. Published at PODC 2026.

Recommended citation: Dinh, T. T. A., Kang, D., & Sadoghi, M. (2026). "FEAT: Fair and Efficient Adversarial Transaction Ordering." In Proceedings of the ACM Symposium on Principles of Distributed Computing (PODC '26) (pp. 489–499). Association for Computing Machinery, New York, NY, USA. DOI: 10.1145/3796701.3815905
Download Paper

FairDAG: Consensus Fairness over Multi-Proposer Causal Design

Published in Proceedings of the VLDB Endowment, 19(2), 265–278, 2025

Order fairness for multi-proposer DAG-based BFT consensus protocols.

Recommended citation: Kang, D., Chen, J., Dinh, T. T. A., & Sadoghi, M. (2025). "FairDAG: Consensus Fairness over Multi-Proposer Causal Design." Proceedings of the VLDB Endowment, 19(2), 265–278. [Preprint]
Download Paper

Brief Announcement: Carry the Tail in Consensus Protocols

Published in Proceedings of the 39th International Symposium on Distributed Computing (DISC 2025), LIPIcs Vol. 356, pp. 59:1–59:7, 2025

Brief announcement on carrying the tail in BFT consensus protocols.

Recommended citation: Gupta, S., Kang, D., Malkhi, D., & Sadoghi, M. (2025). "Brief Announcement: Carry the Tail in Consensus Protocols." In Proceedings of the 39th International Symposium on Distributed Computing (DISC 2025) (LIPIcs, Vol. 356, pp. 59:1–59:7). Schloss Dagstuhl – Leibniz-Zentrum für Informatik. [Preprint] [Blog]
Download Paper

HotStuff-1: Linear Consensus with One-Phase Speculation

Published in Proceedings of the ACM on Management of Data (SIGMOD), 3(3), 1–29, 2025

Linear-communication BFT consensus accelerated with one-phase speculation. Honorable Mention for Best Artifact.

Recommended citation: Kang, D., Gupta, S., Malkhi, D., & Sadoghi, M. (2025). "HotStuff-1: Linear Consensus with One-Phase Speculation." Proceedings of the ACM on Management of Data (SIGMOD), 3(3), 1–29. Honorable Mention for Best Artifact. [Preprint] [Blog]
Download Paper

Spotless: Concurrent Rotational Consensus Made Practical through Rapid View Synchronization

Published in 2024 IEEE 40th International Conference on Data Engineering (ICDE), 2024

Practical concurrent rotational consensus with rapid view synchronization.

Recommended citation: Kang, D., Rahnama, S., Hellings, J., & Sadoghi, M. (2024). "Spotless: Concurrent Rotational Consensus Made Practical through Rapid View Synchronization." 2024 IEEE 40th International Conference on Data Engineering (ICDE), pp. 1916-1929. IEEE. [Preprint]
Download Paper

Preprints

Hermes: Low Tail-Latency Via Prefix Consensus

Published in arXiv preprint arXiv:2607.25916, 2026

A rotating-leader consensus protocol where expired views can still finalize their heaviest common prefix, reducing tail latency without sacrificing safety. A timely honest leader finalizes in 2δ.

Recommended citation: Ranchal-Pedrosa, A., Kang, D., Giridharan, N., Malkhi, D., Sadoghi, M., & Marsh, B. (2026). "Hermes: Low Tail-Latency Via Prefix Consensus." arXiv preprint arXiv:2607.25916.
Download Paper