This repository contains my implementation of various Generative AI projects based on the YouTube playlist "Generative AI from Basic to Advance" by Sunny Savita. The playlist is dedicated to learning Generative AI step by step, covering foundational concepts, LangChain, LlamaIndex, RAG, and advanced AI pipelines.
From these tutorials, I have implemented:
- Generative AI basics: text generation, embeddings, and vector search
- LangChain and LlamaIndex integration for RAG pipelines
- Querying and retrieving information from vector databases (MongoDB Atlas, FAISS)
- Using embeddings and advanced search techniques for QA over documents
- Incorporating multi-modal inputs like images in RAG pipelines
- Fine-tuning and leveraging Google Gemini API for LLM responses
- End-to-end RAG applications combining document retrieval, embeddings, and LLM generation
- API Choice: I have used Google Gemini API instead of OpenAI API. Google Gemini API is free, whereas OpenAI API is a paid service.
- Package Updates: Most of the LangChain and LlamaIndex code from the tutorials has changed over time due to package updates and deprecations. I have adapted those changes in my code. Future updates may require further modifications.
- References: All tutorials and guidance are credited to Sunny Savita’s YouTube playlist: Generative AI from Basic to Advance
Note: The
.gitignoreis configured to exclude temporary and sensitive files likelogs/,__pycache__/,.env, and virtual environments (rag_qa_gemini_llama_venv) to keep the repo clean.
- Clone this repository:
git clone https://github.com/prakadeesh01/genai_tutorial.git
cd genai_tutorial- Install dependencies (suggested in a virtual environment):
pip install -r requirements.txt-
Set up your environment variables (.env) with your Google Gemini API key.
-
Open the notebooks in Colab or Jupyter and run each cell sequentially.
This repository is based on tutorial guidance but contains my adaptations.
Feel free to submit improvements, fixes, or updates for package compatibility.
This project is licensed under the MIT License.
See the LICENSE file for details.