I build backend systems in Go: event pipelines, the infrastructure they run on, and software that reaches out and moves real hardware.
Go services that talk through Kafka and gRPC, built for high throughput. Most of the work is in the parts nobody sees: backpressure, idempotent consumers, and knowing exactly what happens on a retry.
Kubernetes on AWS EKS, shipped through GitOps with Argo CD. Changes go through a pull request, and a rollback is a revert.
Embedded Linux, MQTT and small devices at the edge. It's my favourite kind of bug, because you can watch it happen on a real board.
A Go SDK for controlling a Raspberry Pi and wiring it into MQTT.
Most GPIO libraries assume your Go process runs on the Pi. I wanted to write the service on my laptop, drive the board over SSH while I worked on it, and then ship the same code to the Pi. PIoneer does both, and publishes pin changes to MQTT so the rest of a system can react to them.
There's no clever fix for a bad backhand, only a few hundred more of them.
If you're building something that has to scale, or has to touch hardware, I'd like to hear about it.