AWS Builder Center

Builder Center

BinWise - No Waste Goes To Waste: Building an AI Circular Waste System on AWS with Strands Agents SDK

BinWise - No Waste Goes To Waste: Building an AI Circular Waste System on AWS with Strands Agents SDK

How I built BinWise, an AI-powered waste segregation platform that classifies waste and makes India zero-waste, using AWS EC2, S3, IAM, and open-source Strands Agents SDK from DTU Bharat Builds Tour.

Harsh Kumar

Full-Stack & AI/ML Developer | Co-Founder @ Judikit

Hello Builders! I'm Harsh Kumar from Bennett University.

In India, we generate 62M+ tons of waste yearly, but 70% goes to landfill because we throw everything in one bin. The real problem is not waste, it's segregation.
So I built BinWise - No Waste Goes To Waste.

What is BinWise?

BinWise follows a circular economy model:
  1. AI WasteScan: User uploads waste image. My EfficientNetB0 model trained on 1000+ images from Kaggle Waste Dataset (Biodegradable, Recyclable, E-Waste) classifies it in <2 seconds with 93%+ accuracy.
  1. Circular Action: If Biodegradable -> converts to organic fertilizer. If Non-Biodegradable -> connects to verified recycler network and CPCB-certified E-Waste centers with live directions.
  1. AI Reuse Assistant: Before throwing, it suggests creative reuse ideas. Powered by LLaMA LLM.
  1. Impact Page: My favorite feature. Users earn 25 Green Karma Points per scan, track money earned from scrap, CO2 saved, compost generated, and compete on Delhi Leaderboard (DTU vs NSUT vs Bennett).
Tech Stack: React + Vite frontend, Node.js + Express backend, 【entity-MongoDB¦canonical_name=MongoDB】, TensorFlow + Flask API for AI.

How I Used 【entity-AWS¦canonical_name=AWS】 - My Builder Journey

I wanted BinWise to be production-ready, not just a college project. 【entity-AWS¦canonical_name=AWS】 made it possible.
【entity-AWS¦canonical_name=AWS】 Services I Used:
  1. 【entity-AWS¦canonical_name=AWS】 EC2: Hosted my entire full-stack app on a t2.medium EC2 instance. Frontend, backend, and Flask AI API all run live 24/7. Both training and inference happen on EC2.
  1. Amazon S3: Primary storage for all user-uploaded waste images. This keeps 【entity-MongoDB¦canonical_name=MongoDB】 light and makes image retrieval super fast.
  1. 【entity-AWS¦canonical_name=AWS】 IAM: Created secure IAM roles for EC2 to access S3 without hardcoding keys.
  1. Future with 【entity-AWS¦canonical_name=AWS】 Bedrock: Planning to move my LLaMA Reuse Assistant to Bedrock for serverless LLM inference.
AWS Open Source Stack I Used (Learnt at DTU Workshop):
  1. Strands Agents SDK: This was taught in the DTU Bharat Builds session! I used Strands Agents SDK to build my AI Reuse Assistant agent that understands waste context and gives reuse ideas.
  1. AWS CDK: Used CDK to define my EC2 + S3 infra as code.
  1. Boto3: Used Boto3 Python SDK to connect my Flask API with S3.
I also deployed my model on EC2 and learned how to make a scalable architecture that can handle 1 lakh users.
In short: EC2 for Compute, S3 for Storage, IAM for Security, and Strands Agents SDK + CDK + Boto3 from AWS Open Source.
BinWise is now live on AWS and ready for Municipal Corporations and Smart Bins integration. Because waste is not waste, it's a resource.
Thanks to 【entity-AWS Builder Center¦canonical_name=AWS】 and WeMakeDevs for Bharat Builds Tour!
#【entity-AWS¦canonical_name=AWS】 #BharatBuilds #StrandsAgents #EC2 #S3 #AIForGood #BinWise
Live Demo:  https://www.youtube.com/watch?v=w\_0DNyaLhNg  GitHub:  https://github.com/HarshKumar100/Bharat-Build-AWS-  Video:  https://www.youtube.com/watch?v=w\_0DNyaLhNg 
# artificial-intelligence# sustainability# amazon-s3# amazon-ec2# gen-ai
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article