AI-Powered-Image-Analyzer-on-AWS-using-Terraform

🤖 AI-Powered Image Analyzer on AWS using Terraform, Rekognition & Amazon Bedrock

🧠 Project Overview

🚀 Live Demo:

https://da1nzwfv653xi.cloudfront.net/

This project demonstrates how to design and deploy a serverless AI-powered image analysis application using modern cloud-native architecture and Infrastructure as Code (IaC).

The system allows users to upload an image through a web interface, automatically detects objects using AWS Rekognition, and generates a natural-language description using Amazon Bedrock (Mistral LLM). The entire infrastructure is provisioned using Terraform and automatically deployed using GitHub Actions CI/CD.

This project reflects practical experience in building end-to-end AI-powered cloud applications, integrating AI services, serverless compute, and automated infrastructure deployment.

---

⭐ Key Highlights

-Built a serverless AI application using AWS Rekognition + Amazon Bedrock

-Designed a scalable cloud architecture using API Gateway, Lambda, S3, and CloudFront

-Implemented Infrastructure as Code (IaC) using Terraform

-Automated deployments using GitHub Actions CI/CD

-Deployed a live production demo on AWS CloudFront CDN

---

🎯 Project Goals

-Build an AI-powered image analysis application

-Deploy a serverless architecture on AWS

-Automate infrastructure using Terraform

-Implement CI/CD pipeline using GitHub Actions

-Demonstrate cloud architecture design principles

-Deliver a scalable, low-maintenance AI solution

---

🧰 Technologies Used

☁️ Cloud Services

| Service | Purpose | | —————————- | ——————————— | | Amazon S3 | Hosts the frontend static website | | Amazon CloudFront | CDN for fast global delivery | | AWS Lambda | Backend processing logic | | Amazon API Gateway | Exposes the REST API | | AWS Rekognition | Image label detection | | Amazon Bedrock (Mistral) | AI description generation | | AWS IAM | Role and permission management |

⚙️ DevOps & Infrastructure

| Tool | Purpose | | —————— | ———————- | | Terraform | Infrastructure as Code | | GitHub Actions | CI/CD automation | | Git | Version control |

💻 Programming & Frontend

| Technology | Purpose | | ————————— | —————————— | | Python | Lambda backend | | HTML / CSS / JavaScript | Web interface | | JSON API | Frontend–backend communication |

---

🏗️ Architecture Diagram

Below is the system architecture illustrating how the application processes images and generates AI insights.

🏗️ Architecture Design Principles

The system was designed following modern cloud architecture best practices:

-Serverless-first architecture to minimize infrastructure management

-Decoupled frontend and backend layers

-Infrastructure as Code (IaC) for reproducible deployments

-Event-driven processing using AWS Lambda

-Global content delivery through CloudFront CDN

-Secure service communication via IAM roles

---

🔄 System Workflow

1. User Uploads Image

The user uploads an image through the web application hosted on Amazon S3.

1. CloudFront Delivers Frontend

CloudFront provides:

-global CDN delivery

 -HTTPS support

 -improved frontend performance

1. API Gateway Receives Request

The frontend sends the image as a base64 payload to POST /analyze.

1. Lambda Processes Image

Lambda performs:

  -Base64 decoding

 -Image processing

 -Rekognition API calls

 -Bedrock LLM prompt generation

1. Rekognition Detects Labels

Example detected labels:

-Dog

-Border Collie

-Grass

-Field

-Outdoor

-Sky

1. Bedrock Generates Description

Using the labels, Bedrock generates a natural-language description:

 -A Border Collie running through a green field with yellow flowers under a blue sky.

1. Results Returned to Frontend

The Lambda function returns: { “labels”: […], “description”: “…” }

---

🚀 Application Demo

The application is fully deployed on AWS and accessible through Amazon CloudFront.

👉 Try the application here:

https://da1nzwfv653xi.cloudfront.net/

---

🌍 Live Deployment Architecture

The application is deployed using the following cloud infrastructure:

-Frontend Hosting → Amazon S3

-CDN Delivery → Amazon CloudFront

-API Layer → Amazon API Gateway

-Compute Layer → AWS Lambda

-AI Services → Rekognition + Bedrock

-Infrastructure Provisioning → Terraform

-Deployment Automation → GitHub Actions

---

⚙️ Infrastructure as Code (Terraform)

All infrastructure resources are created using Terraform.

Terraform provisions:

-IAM roles & policies
  
-Lambda function deployment
  
-API Gateway REST API
  
-Lambda permissions
  
-S3 static website hosting
  
-CloudFront CDN
  
-Terraform outputs for endpoints

Example IaC flow:

This ensures:

-reproducible infrastructure
  
-automated deployments
  
-version-controlled cloud resources

---

🔄 CI/CD Pipeline (GitHub Actions)

The project includes a GitHub Actions workflow that automatically deploys infrastructure.

Pipeline workflow:

Benefits:

-automated infrastructure deployment
  
-reproducible environments
  
-DevOps best practices
  
-secure secret management

---

💡 Why This Project Matters

Organizations manage massive volumes of image data but often rely on manual tagging and description generation. This solution demonstrates how AI + cloud infrastructure can automate image understanding and support:

-automated media tagging

-AI-powered product description generation

-accessibility improvements for visual content

-intelligent digital asset management systems

---

🚀 Future Enhancements

Possible improvements include:

-Upload images directly to S3 instead of base64 encoding

-Add authentication using Amazon Cognito

-Implement CloudWatch monitoring dashboards

-Support multiple image uploads

-Optimize LLM prompts for richer descriptions

-Add image storage and retrieval functionality

---

⭐ If you find this project interesting, feel free to star the repository or connect with me to discuss cloud, AI, and data engineering solutions.