P-05Cloud · Team project
Serverless Image Processor
An AWS image pipeline, described as code.
Serverless image processor on AWS with Terraform across three environments and GitHub Actions deployments.
- Year
- 2026
- Type
- Team project
- Languages
- HCL, JavaScript
- Demo
- No public demo
The problem
Asynchronous image processing requires coordinating storage, queues and compute, with reproducible and secure deployments across environments.
The approach
- Uploads through API Gateway, storage in S3 and circular cropping in an SQS-triggered Lambda, with a dead-letter queue and a CloudWatch alarm.
- Terraform modules for dev, qa and prod, each with its own remote state.
- GitHub Actions pipeline with OIDC authentication, so no access keys live in the repository.
My part
My documented responsibilities: bootstrap (remote state, OIDC and deployment roles), pipeline, networking module, and the QA and production environments.
System flow
Drawn with the process flow diagram notation used in chemical engineering.
- Input
- Process
- Storage
- Output