ML inference architecture, applied to agent systems.
Sean Fraser / ML Inference Architecture
More than two decades in networking and security, now applied to inference serving, agent infrastructure, and the controls that keep those systems dependable.
Editorial archive
About this work
This site began as a place to dissect complex cloud and security technologies for practitioners. It now sits inside a broader technical practice spanning ML inference architecture, agent infrastructure, observability, policy enforcement, open models, networking, and security.
The aim remains practical: connect architecture to evidence, explain tradeoffs plainly, and leave behind work that another operator can inspect.
Working domains
Areas of expertise
- 01
ML inference architecture
Serving paths, memory, scheduling, and the limits that decide whether a request meets its objective.
- 02
Agent infrastructure
Identity, control, and the path an agent takes in production.
- 03
Observability
What the system did, from traces and operator evidence.
- 04
Policy enforcement
What an agent may reach, and the boundary around that reach.
- 05
Open and local models
Models you can run, inspect, and serve yourself.
- 06
Networking and security
Paths, failure surfaces, and the controls that keep the rest dependable.
Experience
Professional journey
-
Present
ML Inference Architecture
Designing inference serving paths, memory and scheduling limits, and the controls that keep model serving dependable.
-
Previous role
Network Security Engineering
Building secure network architectures and implementing defense-in-depth strategies for enterprise environments.
-
Career foundation
20+ Years in Technology
From foundational network security principles through contemporary cloud defense methodologies.