Multi-Agent Inventory Intelligence System on AWS Free Tier

I built this end-to-end inventory management system entirely on AWS free tier, running in the Mumbai (ap-south-1) region for low latency and cost control. It processes uploaded CSV spreadsheets (with product details like name, quantity, SKU, etc.) to predict stockouts, prioritize reorders, send alerts, and generate reportS and this all automated and serverless.
How it works:
Trigger options: Upload to S3 bucket ("inventory-uploads") auto-starts via EventBridge, or manual API call through API Gateway.
Orchestration: AWS Step Functions coordinates 7 Lambda functions (originally 5 agents, split for better modularity):
Ingestion: Reads CSV from S3, stores in DynamoDB ("inventory-item-table").
Analytics: Predicts stockouts/demand (using SageMaker for forecasting).
Optimizer: Sets reorder priorities (high/medium).
Alerts: Sends prioritized emails via SNS.
Reporting: Generates final report to S3 ("inventory-agent-outputs").
Parallel execution for alerts/reporting to speed things up.
Data storage: 3 DynamoDB tables (items, forecasts, reports history); S3 for inputs/outputs/ML artifacts; optional static website hosting.
Security: Custom IAM roles for Lambda, Step Functions, and SageMaker—ensuring services only access what they need.

This project taught me production-ready cloud design—fully automated, observable, and cost-optimized. Feedback welcome!

hashtag#AWS hashtag#Serverless hashtag#Lambda hashtag#StepFunctions hashtag#DynamoDB hashtag#S3 hashtag#VPC hashtag#EC2 hashtag#AmazonEventBridge hashtag#APIGateway hashtag#CloudWatch hashtag#DataEngineering hashtag#CloudArchitecture

Видео Multi-Agent Inventory Intelligence System on AWS Free Tier канала Learn with Menush