Author
Kim
Backend software engineer working with Java, Spring Boot, and Kafka. I write about engineering problems, performance work, and architecture decisions from development and operations.
-
JSON to Java, Kotlin and TypeScript in the Browser: What JSON2Class Infers and Where It Stops
A walk through the type inference behind JSON2Class, a browser-only JSON to class generator I build and maintain: the Java, Kotlin and TypeScript it emits for one order payload, the annotation directives you can write inside the JSON, how array element types widen across every element, and the limits that remain before you paste production…
-
Temperature Scaling Under Data Drift: What Recalibration Fixes and What It Cannot
A reproducible NumPy lab: temperature scaling never changes accuracy, a temperature fitted on clean data stops working after covariate drift, and under concept drift the best temperature runs away because the model is wrong.
-
Canary Releases With Automated Rollback on Kubernetes: Flagger and Istio, Run and Recorded
A Flagger v1.45.0 + Istio 1.31.0 canary on kind: a healthy candidate promoted in five weight steps, a candidate returning HTTP 500 rolled back after five failed checks, the VirtualService weights at every step, and the metric-window mistake that almost rolled back the healthy one.
-
Kafka Security with Mutual TLS and ACLs: A KRaft Lab, the Principal Trap, and the Commands That Still Exist
A single-node Kafka 4.3.1 broker with mutual TLS on every listener and ACLs, tested with eight client scenarios: what the authorizer sees as your identity, why the ZooKeeper-era ACL commands fail, and what each failure looks like.
-
Server-Sent Events with Spring Boot WebFlux: A Tested Implementation and Its Limits
A Spring WebFlux SSE endpoint built with Reactor Sinks, verified in Chrome: named events, automatic reconnection, Last-Event-ID replay, and the sink behaviors that break shared streams.
-
Async Logging with the LMAX Disruptor in Java: What Blocks, What Drops, and What You Lose
A tested Disruptor 4.0.0 async logger: why ringBuffer.next() blocks when the buffer is full, how tryNext() drops, two ways to lose log lines silently, and a benchmark with stated boundaries.
-
Shadow Testing an AI Model: The Metrics, and the Two Ways the Shadow Lies
A reproducible NumPy lab that scores a production and a candidate model on the same request stream, computes the paired comparison a shadow run should produce, and shows two cases where the shadow verdict is wrong: label delay under a traffic-mix shift, and labels that only exist for what production approved.
-
Blue-Green Deployments in Kubernetes: What a Service Selector Switch Actually Does, Measured
Blue and green Deployments, a Service selector switch under a 50 ms request loop, readiness gating, a failure-then-revert sequence, and the merge-patch trap, all recorded on kind with Kubernetes 1.34.11.
