DEV Community

JPJP19
JPJP19

Posted on

Introducing WasteConnect — AI-Powered Recycling Marketplace

#ai #programming #productivity #api

Introducing WasteConnect — AI-Powered Recycling Marketplace

We built WasteConnect, an intelligent recycling marketplace that connects waste generators with suitable recyclers using AI-powered waste understanding, semantic matching, and competitive bidding.

What does WasteConnect do?

WasteConnect turns the process of selling recyclable waste into an intelligent marketplace.

  1. AI Waste Identification
    Users upload a waste image and optional description. Gemma analyzes the waste and identifies its category, materials, condition, and potential hazards.

  2. Automatic Listing Creation
    Instead of manually filling multiple fields, AI converts the uploaded information into a structured waste listing that the user can review and confirm.

  3. Intelligent Recycler Matching

    The system identifies recyclers who can actually handle the material. Semantic embeddings + pgvector** match the waste with recycler capabilities even when different terminology is used.

  4. Smart Eligibility Filtering

    Deterministic algorithms check important constraints such as material compatibility, service radius, recycler verification, and safety authorization.

  5. Competitive Bidding
    Multiple eligible recyclers can compete for the same waste by submitting bids, creating a transparent marketplace instead of a single fixed-price transaction.

  6. AI-Powered Bid Analysis
    Gemma analyzes competing offers using factors such as price, distance, pickup convenience, reliability, and risk, helping the user understand which option provides the best overall value.

  7. Recycler AI Copilot**

    Recyclers can interact with the platform using natural language to discover relevant waste listings and identify promising opportunities.

  8. Trust & Dispute Intelligence**

    Transaction history and quantity discrepancies can be analyzed to identify potential issues and assist in resolving disputes.

How we built it

The core idea was to give each component the responsibility it is best suited for:

Gemma 4 + Ollama** → vision, reasoning, extraction, reranking and bid analysis

  • 🔎 Sentence Transformers → semantic embeddings
  • 🗄️ PostgreSQL + pgvector → semantic recycler search
  • ⚙️ FastAPI + Python → backend and APIs
  • 💻 Next.js + React + Tailwind CSS → frontend
  • 🐳 Docker → development and deployment
  • 📐 Deterministic algorithms → distance, filtering, validation and business rules

🔥 What makes it different?

WasteConnect is not simply an "upload → ask AI → get an answer" application.

It combines:

Multimodal AI + Semantic Search + Vector Database + Deterministic Algorithms + Competitive Marketplace + AI Decision Support

Gemma understands and reasons, embeddings discover semantic relationships, algorithms enforce correctness, and the database provides real-world context.

The final decision always remains with the user — AI helps them make a better-informed decision.

Top comments (0)