GenAI Use Cases

This repository contains practical Generative AI (GenAI) projects and pipelines for real-world use cases.

Each folder in /usecases is a standalone use case, complete with code, setup instructions, and documentation.
The /examples folder contains sample programs I created to learn and experiment with technologies like LangChain.

1. Project Overview

This repository showcases multiple Generative AI use cases.

Current example:

Internal Documentation RAG Pipeline:
- Ingests and cleans Markdown/HTML documentation
- Generates text embeddings (OpenAI or HuggingFace)
- Stores and searches embeddings in ChromaDB
- Retrieves and summarizes content with LLMs using LangChain

More use cases will be added over time.

2. How to Run the Internal Doc Jupyter Notebook

A. Clone the Repository

git clone https://github.com/rguthaa/genai-usecases.git
cd genai-usecases/usecases/internal_doc_pipeline

B. Launch Jupyter Notebook

If you don’t have Jupyter installed, run:

pip install jupyter

Or, in a virtual environment:

source venv/bin/activate
pip install jupyter

Now launch:

jupyter notebook

C. Environment Setup

The first cell in each notebook runs the setup script for you.
If you run it manually, you must specify the model provider (OpenAI, VertexAI, etc):

# For OpenAI
bash setup.sh OpenAI

# For Google Vertex AI
bash setup.sh VertexAI

The setup script will automatically call the appropriate provider-specific setup (like vertex_ai_setup.sh) based on the argument you provide.

D. Follow the Notebook

Step through the cells to load data, clean, embed, search, and generate LLM-based answers.

3. How to Run the Internal Doc Pipeline as a Python Project

A. Clone the Repository (if not already done)

git clone https://github.com/rguthaa/genai-usecases.git
cd genai-usecases

B. Setup Environment

bash setup.sh OpenAI      # or VertexAI, depending on your provider

C. (Optional) Google Vertex AI Setup

If you use Vertex AI, the setup script will handle this step for you.

D. Run the Pipeline

cd usecases/internal_doc_pipeline/src
python main.py 1

The script will process the docs, create embeddings, run retrieval, and summarize with your chosen LLM.

Environment Variables

You must configure environment variables for API keys and service credentials. A .env.example file is provided in the repo root.

To use it:

Copy .env.example to .env in the root directory:
```
cp .env.example .env
```
Edit .env and fill in your actual keys/paths, for example:
```
OPENAI_API_KEY=your-openai-api-key
```
Do NOT commit your .env file to git. .gitignore includes .env for your safety.

Directory Structure

usecases/ internal_doc_pipeline/ notebooks/ # Jupyter notebooks for experimentation src/ # Core Python source code for the pipeline setup.sh # Main setup script (pass model provider as argument) vertex_ai_setup.sh # GCP/Vertex AI setup (called as needed) examples/ ... # Sample programs, LangChain experiments, etc.

Other usecase folders may be added in the future

Requirements

Python 3.9+
(Optional) Google Cloud CLI for Vertex AI workflows
OpenAI API key and/or HuggingFace token (for embeddings)
.env file with your API keys and credentials (see above)

More Information

Related Medium Article: RAG Pipeline for Summarizing Internal Documentation

Contributing and Questions

Questions, issues, and pull requests are welcome. If you are experimenting with similar GenAI pipelines, feel free to fork, extend, or reach out via the Issues tab.

hackur/genai-usecases