Pallet
Two infra engineers who turned a 3 a.m. on-call nightmare into a quiet pager.
The situation
Pallet runs inventory software for small e-commerce brands. They'd grown fast, and their infrastructure had grown by accretion — nobody owned it, deploys were scary, and the on-call engineer averaged two middle-of-the-night pages a week. People were getting tired in a way that turns into resignations.
They needed real DevOps experience, not a developer who'd watched a Kubernetes tutorial. And they needed it without a months-long hiring process they didn't have the runway for.
What we did
We matched two engineers with deep AWS and reliability backgrounds, both with strong overlap on Pacific hours so they could pair with the existing team in real time rather than over async lag.
Their first month was deliberately unglamorous: fix the noisy alerts, document the deploy process, and put guardrails on the parts of the system that kept breaking. Only once the bleeding stopped did they move on to the longer-term platform work.
I can't overstate what it does for a team to just… sleep through the night again. They didn't show up and rewrite everything. They fixed the stuff that was actually hurting us, in order, and explained it as they went.
How it turned out
Within two months the pager went quiet. Deploys went from a dreaded event to a non-event. Pallet's existing engineers, who'd been one bad month from burnout, stayed. The two infra goblins are still embedded with the team almost a year later.