Blog
Technical writing on distributed systems and infrastructure
Infrastructure as a Library
Why "Infrastructure as Code" got the name wrong. Terraform treats infrastructure as a passive list with reconciliation loops. True dynamic infrastructure should work like a library: borrowed, scoped, automatically returned via RAII.
The Banker's Algorithm for Distributed Systems
How classical OS theory informs distributed systems design. Dijkstra's 1965 deadlock avoidance strategy provides the foundation for Observability-Driven Development: systems that observe their entire state space before acting.
Why Stateless Failover: An Engineering Deep Dive
We evaluated OS-level container checkpointing (CRIU) for warm migration on GPUs. Why we rejected it and chose stateless failover instead.