PO Classification System – LLM Powered 📌 Project Overview

PO Classification System is an AI-driven application that automatically classifies Purchase Order (PO) descriptions into predefined categories using a Large Language Model (LLM). Instead of traditional ML training, the system relies on prompt-based reasoning via the Groq API.

🚀 Problem Statement

Organizations process large volumes of purchase orders. Manual classification is:

• Time-consuming • Error-prone • Difficult to scale

An intelligent automated solution is needed to improve efficiency and accuracy.

✅ Solution

This project implements an LLM-based classification system that:

• Accepts PO descriptions from users • Constructs optimized prompts • Sends requests to the Groq LLM endpoint • Returns predicted classification labels

No dataset training or feature engineering required.

🛠️ Tech Stack

Language • Python

Frontend / UI • Streamlit (interactive web interface)

AI / Model Backend • Groq API • Model: openai/gpt-oss-120b

Core Libraries • streamlit • groq

🏗️ Project Architecture

• UI Layer → app.py (Streamlit interface) • Logic Layer → classifier.py (classification workflow) • Prompt Management → prompts.py • Category Definitions → taxonomy.py • Configuration → settings.py

⚙️ How It Works

User enters a PO description

Input is processed and validated

Prompt is dynamically generated

Prompt sent to Groq API

LLM predicts classification category

Result displayed in UI

✨ Key Features

✔ LLM-powered classification ✔ No model training required ✔ Lightweight & fast ✔ Clean modular structure ✔ Easily extendable taxonomy

🎯 Use Cases

• Procurement automation • Enterprise workflow optimization • Intelligent document processing • AI-based categorization systems

📦 Installation & Setup

1️⃣ Clone Repository

git clone cd PO_Classification

2️⃣ Install Dependencies

pip install -r requirements.txt

3️⃣ Configure API Key

Add your Groq API key inside Streamlit secrets:

.streamlit/secrets.toml

Example:

GROQ_API_KEY = "your_api_key_here"

4️⃣ Run Application

streamlit run app.py 📈 Future Improvements

✔ Add file / PDF input support ✔ Improve prompt optimization ✔ Add logging & analytics ✔ Deploy as API service ✔ Multi-label classification

👨‍💻 Author

Developed for learning, experimentation, and AI-based automation exploration.