ClassiCore is an enterprise-ready tool that intelligently classifies product/service descriptions into accurate UNSPSC codes using advanced semantic search and AI inference.

---

📌 Table of Contents

• System Architecture
• Tech Stack
• Deployment Prerequisites
• Step-by-Step Deployment on AWS EC2 (g5 GPU)
• Running the Application
• Directory Structure
• API Contract
• Contributors

---

🏗️ System Architecture

                                         ┌────────────────────┐
                                         │   React Frontend   │
                                         └────────┬───────────┘
                                                  │
                                                  ▼
                                         ┌────────────────────┐
                                         │  Spring Boot API   │
                                         └────────┬───────────┘
                                                  │
                                                  ▼
                                      ┌──────────────────────────┐
                                      │  FastAPI AI Classifier   │
                                      └────────┬───────────┬─────┘
                                               │           │
                                ┌──────────────┘           └───────────────┐
                                ▼                                          ▼
                       SQLite (UNSPSC Data)                  SentenceTransformer (Embeddings)

---

💻 Tech Stack

Layer	Technology
Frontend	React.js, Vite, Tailwind
Backend API	Spring Boot (Java)
AI Service	FastAPI (Python), Hugging Face Transformers
Model Hosting	Mistral / IBM Granite
Embeddings	SentenceTransformers (all-MiniLM)
DB	SQLite (embedded UNSPSC dataset)
Infra	AWS EC2 g5.xlarge (with GPU), Docker
Logging	Console + future plan for ELK/CloudWatch

---

⚙️ Deployment Prerequisites

1. AWS EC2 g5.xlarge instance (Ubuntu 22.04)

2. Security Group:

   • Port 3000 (Frontend)
   • Port 8080 (Spring Boot Backend)
   • Port 8000 (Python FastAPI)

3. Install Docker

4. Install NVIDIA GPU drivers:

   sudo apt update && sudo apt upgrade
   sudo apt install nvidia-driver-525
   reboot

5. Install NVIDIA Container Toolkit:

   sudo apt install nvidia-container-toolkit
   sudo systemctl restart docker

6. Test GPU with Docker:

   docker run --gpus all nvidia/cuda:11.0-base nvidia-smi

---

🚀 Step-by-Step Deployment on AWS EC2 (g5 GPU)

1. Clone Project

git clone https://github.com/your-org/classicore.git
cd classicore

2. Build and Run with Docker Compose

docker compose --profile prod up --build -d

This spins up all 3 services: frontend, backend (Spring Boot), and ai-classifier (FastAPI) with GPU access enabled.

3. Check Logs

docker logs classicore-backend
docker logs classicore-classifier
docker logs classicore-frontend

---

🖥️ Running the Application

• Access Frontend: http://<EC2-IP>:3000
• API (optional): http://<EC2-IP>:8080
• AI Classifier: http://<EC2-IP>:8000

---

📁 Directory Structure

classicore/
├── frontend/              # React.js UI
├── backend/               # Spring Boot API
├── ai-classifier/         # Python FastAPI + Embeddings
├── unspsc.sqlite          # Classification DB (auto-mounted)
├── docker-compose.yml     # Multi-container setup with GPU
├── README.md              # This file

---

📨 API Contract

POST /classify

{
  "product_name": "medical face mask"
}

Response:

[
  {
    "code": "42131728",
    "label": "Antifog surgical masks and tie back masks",
    "confidence": 66.7,
    "source": "Semantic Search"
  },
  ...
]

---

👨‍💻 Contributors

• Shakeel – Architect & Full Stack Developer LinkedIn • Email

📚Java|Spring Boot|AWS|React|Microservices|CI/CD|Semantic Search|AI Inference
🧑‍💻A decade of experience scaling cloud platforms & leading global agile teams to boost performance.
💻Passionate about building intelligent systems that bridge AI and enterprise software.

---