Backend-leaning engineer who spends most of my time around AI agents and the infrastructure that keeps them running. Got pulled into the LLM-ops side mostly out of necessity and ended up finding the orchestration and observability problems more interesting than the models themselves. Still figuring out how much of the "agent" hype survives contact with production.