🔗 Repo: github.com/Push/terraform-aws-labs/day18

🎯 Goal: Build a fully automated image pipeline — upload → process → 5 variants — zero servers, zero manual steps.

✅ You’ll build:

• ✅ Upload S3 bucket (source) + Processed S3 bucket (destination)
• ✅ Lambda function (Python + Pillow) — compress, resize, convert formats
• ✅ S3 event trigger — auto-invoke Lambda on s3:ObjectCreated:*
• ✅ Least-privilege IAM roles — no S3FullAccess!
• ✅ CloudWatch logging — track cold starts, errors, durations

🧠 Why Serverless? (The “It Works on My Machine” Fix)

💡 Golden Rule:

“If your workload is event-driven and <15 mins — Lambda is your friend.”

📦 Architecture Diagram

flowchart LR
  A[Upload Image<br/>to S3] -->|s3:ObjectCreated:*| B[Lambda Trigger]
  B --> C[Lambda Function<br/>(Python + Pillow)]
  C --> D1[Compressed JPEG<br/>(85% quality)]
  C --> D2[Low-Quality JPEG<br/>(60% quality)]
  C --> D3[WebP<br/>(85% quality)]
  C --> D4[PNG<br/>(lossless)]
  C --> D5[Thumbnail<br/>(200x200)]
  D1 --> E[Processed S3 Bucket]
  D2 --> E
  D3 --> E
  D4 --> E
  D5 --> E
  C --> F[CloudWatch Logs]

✅ Critical Flow:

1. User uploads photo.jpg to upload-bucket-dev

2. S3 emits s3:ObjectCreated:Put event

3. Lambda invoked with event payload:

   {"Records": [{"s3": {"bucket": {"name": "upload-bucket-dev"}, "object": {"key": "photo.jpg"}}}]}
   
   

4. Lambda:

   • Downloads photo.jpg from S3
   • Processes with Pillow (5 variants)
   • Uploads to processed-bucket-dev

5. Logs duration/memory to CloudWatch

✏️ Hands-On: Terraform Implementation

🔹 File Structure (/day18/)