Spring GDS 25th Anniversary
A logistics company that ships to 190 countries built something to ship to itself.
Serverless is a way of running code without managing the servers it runs on. There are still servers. You just stop owning them. The cloud provider handles capacity, scaling, and patching, and you deploy functions that wake up to handle a request and disappear when they are done. AWS Lambda, Cloudflare Workers, and Google Cloud Functions are the common platforms.
The model changes the economics. You pay per execution and per millisecond of compute, not for a machine sitting idle at 3am. Scaling is automatic, so a function that handles ten requests handles ten thousand without anyone provisioning hardware. An image-processing job that fires only when a user uploads a photo is a clean fit, since the code runs when needed and costs nothing in between. Serverless usually pairs with managed databases, queues, and storage to round out a full system.
The catch is the cold start. A function that has not run recently takes longer to spin up the first time, which matters for latency-sensitive work. Long-running jobs, heavy persistent connections, and very high steady traffic can cost more than a plain server. The right fit is spiky, event-driven, or unpredictable workloads. The wrong fit is a constant high-throughput service that never sleeps.
We build serverless when the workload is bursty or the team should not be babysitting infrastructure. APIs, scheduled jobs, webhooks, and event-driven processing run well as functions, and they slot cleanly into our CI/CD pipelines so a push to main deploys without a release ceremony.
The honest part is that serverless is not a default. It can quietly run up cloud cost on steady high-volume traffic, and cold starts hurt user-facing latency if you ignore them. So we measure before we commit and mix serverless with always-on services where that makes more sense. When it fits, the client gets a system that scales itself, costs little at rest, and frees the team to ship features instead of managing machines.
Wondering if serverless fits your workload, or just inflates the bill? Let's check.
A logistics company that ships to 190 countries built something to ship to itself.
Turning a brand into a working business.
Half a million people. One app. Zero chaos.















