Vahab Jabrayilov
CS PhD Candidate @ Columbia University

I am a Ph.D. candidate in Computer Science at Columbia University advised by Kostis Kaffes, focusing on systems performance and low-latency computing.
My work spans networking, operating systems, and machine learning runtimes, with an emphasis on CPU and GPU virtualization. I have built high-performance software such as Machnet, a kernel-bypass networking framework, and interned at Microsoft Research, where I worked on large-scale infrastructure for data and networking.
I have expertise in C/C++, Rust, CUDA, DPDK, eBPF and Linux kernel programming, and my interests lie at high-frequency trading systems and large-scale AI platforms, where reliability and nano/microsecond-scale performance are critical.
Before starting my Ph.D., I built Replicant, a MultiPaxos-based key-value store for strongly consistent cloud replication during a research internship at Penn State, and also gained experience at EPFL, where I worked on distributed systems reliability and failure analysis.
Education
![]() | Columbia University PhD, Computer Science, 2023-2027 (Expected) | ![]() | Columbia University MSc, Computer Science, 2023-2025 |
![]() | Middle East Technical University BSc, Computer Engineering Engineering, 2019-2023 |
Recent News/Posts
Feb 18, 2025 | Pushing Cloud Systems Toward Bare-Metal Performance |
---|
Selected Publications
- VLDBHoliPaxos: Towards More Predictable Performance in State Machine ReplicationProc. VLDB Endow., 2025
- arXiv
- SoCC
- arXivMultiPaxos Made Complete2024
- arXiv