Feature: Zero-infra serverless execution on AWS Lambda

[FernandoCelmer]

Idea

Ensure dotflow runs seamlessly on AWS Lambda with zero infrastructure beyond Lambda + S3. Provide deployment examples and validate all execution modes.

Motivation

Dotflow's key differentiator is zero-infra execution. A Lambda handler should be as simple as:

def handler(event, context):
    config = Config(storage=StorageS3(bucket="workflows"))
    workflow = DotFlow(config=config)
    workflow.task.add(extract_data)
    workflow.task.add(transform_data)
    workflow.task.add(load_data)
    workflow.start(mode="sequential", resume=True)
    return workflow.result()

No server to maintain. No workers. No dedicated database. Just Lambda + S3.

Goals

• Validate sequential mode works on Lambda without modification
• Document limitations of parallel/background modes on Lambda
• Provide SAM and CDK deployment templates
• Provide example Lambda handler with S3 storage
• Ensure StorageDefault (in-memory) works for stateless one-shot workflows
• Handle Lambda's /tmp ephemeral storage for StorageFile fallback
• Test cold start performance with dotflow import

Deliverables

1. Example project: examples/aws-lambda-sequential/
2. SAM template (template.yaml)
3. CDK example (optional)
4. Documentation page on deploying to Lambda
5. Benchmark: cold start time, execution overhead

Related issues

• Feature: Zero-infra cloud-ready execution (Lambda, Cloud Run, etc.) #74 (Zero-infra cloud-ready execution)
• Feature: AWS integrations (Lambda, Step Functions, S3, SQS, EventBridge) #75 (AWS integrations)
• Feature: Checkpoint-based durable state (resume from failure) #101 (Checkpoint-based durable state)
• Feature: StorageS3 provider for AWS S3 persistence #104 (StorageS3 provider)

[FernandoCelmer]

Implemented in PR #183 — Lambda deployers (base, scheduled, S3, SQS, API Gateway) with dotflow deploy --platform lambda.