DLSU CpE Academic Advising AI

A web-based AI agent that provides academic advising for De La Salle University's Computer Engineering program, powered by Pinecone vector database for semantic search.

Features

• Interactive web interface for asking questions about CpE courses
• Powered by a CodeAgent with RAG (Retrieval-Augmented Generation)
• Pinecone Vector Database for semantic search and scalability
• Beautiful, responsive UI with gradient design
• Real-time chat interface
• Fast similarity-based course lookup

Quick Start

Prerequisites

• Python 3.8+
• Pinecone account (free tier available)
• Hugging Face API token

Setup

1. Install dependencies:

   pip install smolagents langchain langchain-core langchain-community flask pinecone-client sentence-transformers

2. Set up Pinecone:

   • Create a free account at https://www.pinecone.io/
   • Create an index named cpe-curriculum with dimension 384
   • Copy your API key

3. Configure environment variables:

   • Copy .env.example to .env
   • Add your Pinecone API key and Hugging Face token

4. Run the application:

   python Main.py

5. Access the web interface:

   • Open http://localhost:5000 in your browser

Usage

• Type your question in the input field (e.g., "What are the prerequisites for THSCP4A?")
• Click "Ask the Adviser" to get a response
• The AI will search the Pinecone vector database for the most relevant courses

Files

• API.py: Agent logic, Pinecone integration, and tools
• Main.py: Flask web application with HTML/CSS frontend
• checklist.json: Curriculum data (automatically loaded into Pinecone)
• PINECONE_SETUP.md: Detailed Pinecone setup guide
• .env.example: Environment configuration template

Architecture

User Query
    ↓
Embedding Generation (all-MiniLM-L6-v2)
    ↓
Pinecone Vector Search
    ↓
Top 3 Similar Courses Retrieved
    ↓
LLM Agent Processes Results
    ↓
Response to User

Technologies Used

• Backend: Python, Flask, SmolAgents, LangChain
• Vector Database: Pinecone
• Embeddings: Sentence-Transformers (all-MiniLM-L6-v2)
• Frontend: HTML5, CSS3, JavaScript
• AI Model: Meta Llama 3.3 70B via Hugging Face Inference API

Benefits of Using Pinecone

✅ Semantic search based on meaning, not just keywords
✅ Highly scalable for large course catalogs
✅ Fast query response times
✅ Managed service (no infrastructure to maintain)
✅ Supports metadata filtering

For Detailed Setup Instructions

See PINECONE_SETUP.md for complete configuration guide.