RESEARCH & ENGINEERING

AI inference. Global networks.
Internet-scale insight.

My work and interests center on AI inference optimization, routing, global network operations and Internet measurement—understanding how distributed systems perform, and making them work better.

AI inference optimization & routing

I work on the connectivity and traffic steering that support distributed inference at Gcore. I’m interested in how network paths, compute placement and serving decisions come together to shape latency, capacity and the user experience.

RELATED GCORE ENGINEERING

Global Inference Routing ↗︎
NVIDIA Dynamo on Gcore ↗︎

These publications describe Gcore’s broader engineering work, including distributed inference and cache-aware request routing. They provide context for the infrastructure I support; they are not presented as papers I authored.

Global network operations & traffic engineering

My work spans peering, backbone architecture, traffic steering and the operation of networks serving CDN, cloud and AI workloads. The focus is practical: choosing good paths, using capacity effectively and maintaining resilient connectivity across regions.

I designed and procured Gcore’s global private backbone. My article on building the backbone explores the architecture and traffic engineering behind it.

Internet measurement

I’m passionate about measuring how the Internet behaves: latency, reachability, routing changes and performance across networks. Measurement connects architecture decisions with what users actually experience.

My recent Hurricane Isaias study uses two independent measurement points to track Internet reachability and recovery across states, cities and networks. The interactive charts show how statewide averages can conceal much deeper losses in coastal communities.

Measuring Hurricane Isaias through Internet reachability →︎

Read my field notes on networks and distributed inference →︎

More writing and technical contributions →︎