I build backends that ship - and this site runs on one of them.

In progress - a low-latency, many-to-many live video platform with map-based geo-discovery, built on a WebRTC SFU.

An in-progress platform for live video at scale. On the backend I am building the media and services layer: a WebRTC SFU that fans out low-latency broadcasts to many concurrent viewers, server-side recording and HLS pipelines for replay, and a geo-spatial, map-based discovery system that places live streams on an interactive map in real time over WebSockets. It ships on iOS, Android and web - TypeScript across a Node.js API, native iOS and Android apps (React Native) and a web client, plus a robust web-based administration portal for operations, moderation and configuration - all backed by Postgres, with containerized media servers and RTMP ingest for external broadcasters. The interesting work has been in real-time media: connection resilience and reconnection, video orientation and codec correctness across devices, and keeping viewer fan-out performant as concurrency grows.

A hands-on SRE lab: full-stack observability on Kubernetes with Dynatrace - distributed tracing, problem detection and incident investigations.

A self-directed lab where I stood up a production-style stack (Node.js/Express services and PostgreSQL) on a Kubernetes cluster (k3s) and instrumented it end to end with Dynatrace - deploying the Dynatrace Operator and OneAgent alongside an OpenTelemetry Collector for automatic APM, distributed tracing, metrics and logs. I then generated realistic incidents (memory leaks, database failures, request timeouts, CPU saturation) and worked the investigation flows end to end: service dependency mapping, latency analysis, problem detection and root-cause analysis. It documents how I approach production observability and SRE incident response.