CareerPath AI (CPAI)

CareerPath AI is an intelligent career advisory engine designed to analyze user career goals, identify skill gaps, and recommend personalized learning paths. By leveraging Large Language Models (LLMs) and Vector Search, the system provides context-aware guidance and curates relevant courses from major educational platforms.

System Architecture & Technical Stack

1. Data Ingestion (ETL Pipeline)

The system employs a robust ETL (Extract, Transform, Load) pipeline to aggregate course data from multiple sources.

• Extraction: Automated scrapers and API connectors collect metadata (titles, descriptions, syllabi) from platforms such as Coursera, Udemy, and localized providers.
• Transformation: Data cleaning and normalization processes standardize course formats, filter low-quality entries, and detect language (TH/EN).
• Loading: Processed data is embedded and stored in a vector database for efficient semantic retrieval.

2. Core AI Engine (RAG - Retrieval Augmented Generation)

The heart of the application is the SkillEngine, which utilizes a RAG architecture:

• Intent Analysis: Google Gemini 2.5 Flash analyzes user input to extract current roles, target career paths, and specific constraints (e.g., preference for free courses).
• Semantic Search: sentence-transformers generate high-dimensional embeddings for both user queries and course catalog data.
• Vector Retrieval: ChromaDB performs similarity searches to find courses that mathematically match the identified skill gaps.
• Contextual Recommendation: The LLM synthesizes the retrieved course data with the user's career context to generate a coherent, step-by-step learning roadmap.

3. Memory & Context Management

• Short-term Memory: Implements LangChain's RunnableWithMessageHistory to maintain conversational context within a session. This allows the system to handle follow-up questions and refine recommendations based on previous interactions.

Installation & Setup

Prerequisites

• Python 3.12+
• UV Package Manager (recommended) or Pip

Configuration

1. Clone the repository.
2. Create a .env file in the root directory.
3. Add your Google API Key:

   GOOGLE_API_KEY=your_api_key_here

Dependency Installation

uv sync

Usage

1. Data Pipeline Execution

Before running the main application, ensure the vector database is populated:

uv run src/engine/vector_manager.py

This command triggers the incremental update process, processing CSV datasets and updating the ChromaDB vector store.

2. Running the Advisor CLI

Start the interactive command-line interface:

uv run main.py

Project Structure

careerpath_ai/
├── data/                   # Raw CSV datasets
├── src/
│   ├── engine/
│   │   ├── skill_engine.py    # Core logic (LLM + RAG)
│   │   └── vector_manager.py  # Vector DB management & ETL
│   ├── model/              # Data models and schemas
│   ├── utils/              # Helper functions (Logging, etc.)
│   └── config.py           # Configuration management
├── vector_store/           # Persisted ChromaDB data
├── main.py                 # CLI Entry point
└── pyproject.toml          # Dependency definitions

License

MIT License