Hire AWS Developer

ECS on Fargate

Our default for most production workloads. Containerized, no servers to patch, scales cleanly. Most client builds land here.

Lambda

Event-driven work, async jobs, webhooks, glue code between services. Where we wouldn't use it: long-running jobs, anything stateful, or workloads where cold starts will hurt the user.

EC2

When we need predictable, sustained compute or a workload that doesn't fit a container model cleanly. Rarer than it used to be. Still the right call sometimes.

RDS (Postgres)

Our default for transactional, structured data. Multi-AZ in production, read replicas when the read pattern justifies it.

DynamoDB

High-throughput, predictable-access patterns. Session stores, audit logs, anything where the access pattern is known on day one. Where it bites: ad-hoc queries you didn't plan for.

S3

Everything that isn't a row in a database. Uploads, backups, static assets, data lake feeds. Lifecycle policies from day one — not month six.

CloudFront

CDN for static assets and read-heavy API caching. Paired with S3 for frontends, with API Gateway for cacheable endpoints.

WAF

Managed rule sets first, custom rules for what slips through. We tune iteratively after seeing real traffic, not by guessing upfront.

IAM

Least-privilege from day one. We've cleaned up enough wide-open IAM messes inherited from other builds to know what it costs to skip this.