🖼️ Serverless Image Resizing with AWS Lambda, Pillow, and SNS

This project implements an event-driven, serverless image processing pipeline on AWS. Images uploaded to Amazon S3 are automatically resized using AWS Lambda and Pillow, stored in a destination bucket, and followed by an email notification using Amazon SNS.

---

📌 Problem Statement

Applications that accept user-uploaded images often need standardized image sizes for thumbnails, profile photos, or product listings. Handling this on traditional servers increases operational overhead and cost.

This project solves the problem using a fully serverless approach:

• Automatic image resizing on upload
• No servers or background workers to manage
• Instant notification once processing is complete

---

🧠 High-Level Architecture

1. User uploads an image to an S3 bucket
2. S3 triggers an AWS Lambda function
3. Lambda resizes the image using Pillow (via a Lambda Layer)
4. Resized image is saved to a destination S3 bucket
5. Lambda publishes a notification to an SNS topic

This design follows an event-driven, cloud-native architecture.

---

🧰 Tech Stack

• AWS S3 – Image storage and event source
• AWS Lambda – Serverless compute
• Python 3.10 / 3.11 – Runtime
• Pillow (PIL) – Image processing
• Lambda Layer – Dependency packaging
• AWS SNS – Email notifications

---

📁 S3 Buckets

Purpose	Bucket Name
Input images	image-resize-input45
Resized images	image-resize-result45
Backup originals	image-resize-backup45

---

❓ Why Pillow and Lambda Layers

AWS Lambda does not include Pillow by default. Since Pillow depends on native system libraries, it must be packaged separately.

A Lambda Layer is used to:

• Package Pillow correctly for Amazon Linux
• Keep Lambda function code clean
• Reuse the dependency across functions

Lambda automatically mounts layers at:

/opt/python/

---

🛠️ Creating the Pillow Layer (AWS CloudShell)

Pillow must be built on Amazon Linux. Use AWS CloudShell:

mkdir pillow-layer
cd pillow-layer
mkdir python
pip3 install pillow==10.2.0 -t python/
zip -r pillow-layer.zip python

Upload pillow-layer.zip as a Lambda Layer and attach it to your function.

---

🧪 Lambda Function Overview

The Lambda function performs the following steps:

• Reads the uploaded image from S3
• Backs up the original image
• Resizes the image using Pillow
• Saves the resized image to a destination bucket
• Sends a notification using SNS

---

🔍 Key Concepts Used

• Event-driven architecture
• Serverless compute
• In-memory file processing
• Native dependency packaging
• Asynchronous notifications with SNS

---

⚙️ Lambda Configuration

Setting	Recommended Value
Runtime	Python 3.10 / 3.11
Memory	512 MB
Timeout	15 seconds

---

🔐 IAM Permissions

The Lambda execution role requires the following permissions:

{
  "Effect": "Allow",
  "Action": [
    "s3:GetObject",
    "s3:PutObject",
    "sns:Publish"
  ],
  "Resource": [
    "arn:aws:s3:::image-resize-input45/*",
    "arn:aws:s3:::image-resize-result45/*",
    "arn:aws:s3:::image-resize-backup45/*",
    "arn:aws:sns:ap-south-1:ACCOUNT_ID:image-resize-notification"
  ]
}

---

🚨 Common Issues

• No module named PIL – Pillow layer not attached
• _imaging import error – Layer built for wrong Python version
• Lambda not triggered – S3 event notification missing
• SNS publish denied – Missing sns:Publish permission

---

🚀 Real-World Use Cases

• User profile photo processing
• E-commerce product image resizing
• CMS image optimization pipelines
• Media processing workflows with alerts

---

📈 Possible Enhancements

• Multiple output sizes
• Preserve original image format
• Watermarking
• EXIF auto-rotation
• CloudFront CDN integration

---

✅ Summary

This project demonstrates a clean, production-ready approach to serverless image processing on AWS. It combines S3, Lambda, Pillow, and SNS into a scalable, low-maintenance pipeline suitable for real systems.

Built as a learning-focused yet production-aligned implementation.