A curated ledger of Generative AI concepts, frameworks, and production-ready workflows.
This repository contains implementations of Generative AI concepts and workflows, following courses by Krish Naik on Udemy. It serves as a personal learning ledger and reference guide, and it will be updated continuously as new tools and frameworks emerge. It also consists of recent developments and changes to the Langchain framework, and the code which are in the courses and here, may vary. Feel free to star it if you find it useful! ⭐
To explore these projects, you need a solid understanding of Python and an enthusiasm for Generative AI terminology.
Environment Setup:
It is highly recommended to use an isolated virtual environment. You can use conda through installation of Anaconda Navigator but recently uv has been in great demand as well. You can install uv package on your local desktop using the command pip install uv and you are all set!
Using Conda:
## Clone the repository
git clone "https://github.com/your-username/GenerativeAI-Bootcamp.git"
cd GenerativeAI-Bootcamp
## Create and activate a virtual environment (Python 3.11.x or 3.12.x)
conda create -n venv python=3.11.x
conda activate venv
## Install the dependencies
pip install -r requirements.txtUsing uv:
## Clone the repository
git clone "https://github.com/your-username/GenerativeAI-Bootcamp.git"
cd GenerativeAI-Bootcamp
## Create and activate a virtual environment (Python 3.11.x or 3.12.x)
uv venv
.venv\Scripts\activate
## Install the dependencies
uv add -r requirements.txtThe langchain framework in this repository is split into two primary segments to demonstrate the transition from legacy LangChain syntax to the modern ecosystem. These resources are from Udemy's Complete Generative AI Course With Langchain and Huggingface course.
🧱 Langchain-V1: Foundational Concepts; Legacy implementations and foundational concepts of Langchain including basic RAG concepts, retrievers, LCEL, and chatbots from scratch.
📂 Concept |
📝 Description |
|---|---|
| 📖 RAG Introduction | Step-by-step implementation of RAG from data ingestion to vector stores. |
| 🔗 LCEL Architecture | Easily connect components to build complex AI applications. Tech: Prompts Language Models Output Parsers |
| 💬 Message History | Save and manage conversational context dynamically. Tech: RunnableWithMessageHistory SystemMessage |
| 🗄️ Vector Stores | Integrate and retrieve data from databases for LLM workflows. Tech: Chroma FAISS |
🚀 Langchain-V2: Modern & Production-Ready; Modern LangChain architecture, production-ready agents, Streamlit deployments, and advanced projects using GROQ and OLLAMA.
⚡ Project |
💡 Description |
|---|---|
| 🤖 QnA Chatbots | Conversational RAG, Document Search, and SQL-based chatbots. Tech: Agents SQL Toolkit |
| 🔍 LLM Search Engine | Agentic lookup for live web information instead of LLM guessing. Tech: Arxiv Wikipedia DuckDuckGo |
| 📝 Text Summarization | Streamlit app to summarize YouTube videos or Website URLs. Tech: load_summarize_chain PromptTemplate |
| 🧮 MathsGPT | Application to perform complex mathematical calculations accurately. Tech: LLMMathChain StreamlitCallbackHandler |
| 🤗 Huggingface Integration | Invoke open-source models directly using repo IDs and tokens. Tech: HuggingFaceEndpoint |
| 📚 PDF-Query RAG | Advanced document storage and querying using Cassandra. Tech: AstraDB cassIO OllamaEmbeddings |
| 💻 CodeLlama Assistant | Interactive coding assistant interface powered by REST APIs. Tech: Gradio |
| 🎥 YT video to Blog | Conversion of YT transcripts to Blog Structure using CrewAI. Tech: CrewAI YouTubeTranscriptAPI |
| ⚖️ Hybrid Search | Combines keyword and vector search for better retrieval accuracy. Tech: Pinecone HybridSearchRetriever |
| 🕸️ Graph Databases | Knowledge graph creation and querying using Cypher language. Tech: Neo4j GraphCypherQAChain |
| ⚙️ Fine Tuning LLMs | Deep dive into training and optimizing model weights. Tech: LoRA QLoRA Lamini API |
| 🔄 LangGraph | State machine graphs to build robust, multi-agent AI systems. Tech: StateGraph Nodes & Edges |
This part consists of different projects created with the help of Gemini open source models. The idea is to use them using the google.generativeai library, and configure using the GOOGLE_API_KEY. These resources are from Udemy's Building Gen AI App 12+ Hands-on Projects with Gemini Pro course.
🚀 Project |
📝 Description |
|---|---|
| 🤖 LIM-LLM Application | Streamlit apps using Gemini flash models for text and image chat. |
| 💬 Conversational Chatbot | Chatbot application that interacts with an LLM with history using Gemini. |
| 🧾 Invoice Extractor | App to extract data from invoices and answer specific queries. |
| 📚 Multiple PDF Chatbot | Upload multiple PDFs and query seamlessly using FAISS and Gemini. |
| 🔍 Document QnA | Simple RAG application for document-based QnA using the Gemini model. |
| 🗄️ Text-to-SQL-Querying | Utilizes sqlite3 to interact with an SQL database and display query outputs. |
| 📄 Resume ATS System | Analyzes resumes against JDs and provide ATS-compliant feedback. |
| 🥗 Calorie Calculator | Estimate calories from uploaded food images for diet tracking. |
| Generate concise summaries of YouTube video transcripts using Gemini. |