An Alchemy app on AWS is one runtime — usually a Lambda
Function — plus the resources it talks to: tables,
buckets, queues, streams. You wire them together with typed
bindings: call S3.PutObject(bucket) in your code and
Alchemy attaches the matching least-privilege IAM statement to
the function's role. The code is the policy.
New here? Set up credentials first, then deploy your first Lambda.
Pick your runtime
- Lambda — serverless, event-driven functions with public Function URLs. The primary documented path: every resource and event source in this section is shown running on Lambda.
- ECS — long-running containers on Fargate. Alchemy bundles your Effect program into a Docker image and runs it as a Cluster + Service + Task.
- EKS — managed Kubernetes (Auto Mode). Deployments, Jobs, raw manifests, and Helm charts in the same TypeScript program as the cluster — no YAML, no kubectl.
- EC2 — full-control virtual machines, with the complete VPC networking toolkit around them.
Not sure which? See Choosing a runtime.
Data
- DynamoDB — key/value tables with
GetItem/PutItembindings and change-data-capture Streams. - S3 — object storage with
GetObject/PutObjectbindings and bucket event notifications. - RDS & Aurora — managed Postgres/MySQL: the
Aurorahelper stands up the whole cluster in one call, with aConnectbinding or the Data API at runtime.
Messaging & events
- SQS — queues with a
SendMessagebinding and aStream-shaped Lambda consumer. - SNS — pub/sub topics with a
Publishbinding and fan-out to queues and functions. - EventBridge & Scheduler — event buses, rules, and cron/rate schedules that invoke your functions.
- Kinesis — ordered, sharded data streams
with a
PutRecordbinding and the sameStreamconsumer surface.
- Sending & managing email — verify an SES
identity, send from a Lambda with the
SendEmailbinding, and manage identity policies, contact lists, dedicated IP pools, tenants, account settings, and deliverability insights. - Receiving inbound email — SES receipt
rule sets and rules that store mail in S3, fan out to SNS, invoke
a Lambda, or bounce it. Region-limited to
us-east-1,us-west-2, andeu-west-1.
Security & secrets
- Secrets & env — .env values via Config, Secrets Manager when the secret is shared, generated, or rotated.
Observability
- CloudWatch — dashboards and metric alarms, declared in the same Stack as the resources they watch.
Frontend & networking
- Websites — the frontend block: static
sites and built Vite apps on S3 + CloudFront as a single
StaticSiteresource. - VPC & networking — the
Networkhelper and the VPC primitives, for when ECS, EKS, or EC2 needs explicit networking.
What are you building?
| You're building | Reach for |
|---|---|
| An HTTP API | Lambda + Function URL + DynamoDB |
| A typed HTTP API | Effect HTTP API on Lambda |
| A typed API for external clients | Effect RPC on Lambda |
| Drive a MicroVM from a Lambda (internal RPC) | MicroVMs + Schemaless RPC |
| An ML training fleet | HyperPod — Slurm or EKS orchestrated, with task governance |
| A REST API with stages/custom domains | API Gateway |
| Your own domain on a site or API | Custom domains with Route53 + ACM |
| A static site | Deploy a static site on S3 + CloudFront |
| An event pipeline | Kinesis → Lambda |
| Background jobs | SQS + Lambda |
| Scheduled jobs | EventBridge Scheduler → Lambda |
| A Postgres database | RDS & Aurora |
| Kubernetes workloads or Helm charts | EKS |
| Object processing | S3 events |
| Sending transactional email | SES sending — identity + SendEmail binding |
| Inbound email pipelines | SES email receiving → S3 / SNS / Lambda |
| Change data capture | DynamoDB Streams |
| API keys or credentials for a function | Secrets & env |
Where next
- Setup — install Alchemy and connect your AWS account.
- Deploy a static site and Effect HTTP API on Lambda — end-to-end guides for the two most common app shapes.
- Providers reference — generated API docs for every AWS resource Alchemy ships, well beyond the blocks documented here (CloudFront, Route53, IAM, and more).