About
I am a software engineer focused on distributed storage, systems infrastructure, and high-performance data paths.
What I work on
My work is centered around distributed storage and infrastructure: metadata and data paths, caching, consistency, failure recovery, RPC, Linux I/O, networking, and performance engineering.
I am especially interested in systems that sit close to AI workloads, where storage, networking, memory hierarchy, and GPU utilization interact in ways that make traditional infrastructure assumptions less useful.
I am a co-author of SOLO, accepted at the ACM SIGOPS Annual Technical Conference (ATC 2026), with collaborators at USTC and ByteDance.
How I think about engineering
I like working from first principles. For me, understanding a system means being able to explain why its architecture exists, where the trade-offs come from, what fails under pressure, and how those decisions show up in production performance.
I use this site to keep a public record of that work: technical notes, deeper essays, project summaries, and things I have learned while building systems.
Areas of interest
- Distributed storage and file systems
- Linux and systems programming
- C++ concurrency and performance
- RPC and networking
- RDMA and NVMe
- AI infrastructure and storage for GPU workloads
- Distributed consensus and fault tolerance