Hello, I am
Dante
Ye Xiang
Networking & AI inference systems engineer.
Role
Networking & AI Inference
Based in
Seattle, WA
Focus
GPU Clusters, RDMA
Status
Open to new roles
Featured project
NIXL
An open-source, high-performance point-to-point data transfer library for GPU inference clusters, part of the AI-Dynamo project.
- C++
- RDMA
- GPU-Networking
Expertise
- Networking & RDMA EFA, RDMA, DPDK — moving bytes between machines without waking up the CPU.
- GPU Communication NCCL, NIXL, collective operations at cluster scale.
- Distributed Inference Serving large models across GPU clusters — latency, not just throughput.
- Systems Performance Finding the bottleneck nobody instrumented yet.
About
I'm a networking engineer who ended up deep in GPU infrastructure — 8+ years at AWS, most recently building the communication layer that GPU clusters use to train and serve large models.
I care about the systems underneath the systems: how packets move, how GPUs talk to each other, and why things get slow at scale. MS in Electrical & Computer Engineering from UC Davis.
Outside of work I build small things — a finance app, a music theory tutor, a few automation tools — mostly to keep learning by shipping.