Using synthetic client bots to load test WebTransport video calls and discovering hidden bottlenecks before they hit production

When your audio worklet becomes the bottleneck and you have to outsmart the garbage collector to make real-time audio work on low-end Android devices

Real‑time video fails at the edges—jitter, latency spikes, dropped packets. To scale `videocall.rs`, the metrics must be objective and actionable. In two weeks we went telemetry‑first: client instrumentation → NATS → `metrics-api` → Prometheus → Grafana. No wishful thinking. No vanity graphs.

The real story of AI collaboration: where it fails, where it succeeds, and how to actually leverage it for 10x productivity gains.

Client-side RTT election for real-time media streaming and high availability across regions

A staff engineer's airport nightmare that cost $665 in 20 minutes. The reason will make you question everything you thought you knew about software reliability—and why the same mistake nearly caused a nuclear disaster.

Science-backed guide to productive 2 AM coding sessions. Learn why tired brains solve problems better and get practical tips for late-night debugging breakthroughs.

A raw look at backend reliability, observability, and leadership through the lens of extreme ownership from a staff software engineer at May Mobility who architects real-time ETA services for Lyft and Uber integrations.