GenAI Foundry

📢 About GenAI Foundry

This repository was created to make the world of Generative AI accessible, simple, and hands-on.

Our mission:

• 🚀 Empower learners to experiment with LLMs and RAG pipelines.
• 🧠 Provide clear, beginner-friendly tutorials.
• 📚 Foster creativity and exploration with GenAI tools.

Feel free to fork, adapt, and expand these examples! 🎯

---

📥 New Here? Start With This ⬇️

Download the Student Quick Start Guide (PDF)

The Quick Start Guide walks you through everything with screenshots — how to get the code, which setup path to choose, and what you should see when it's working. Download this before doing anything else.

---

🗺️ Learning Path

Follow this sequence for the best experience:

① Tokens & Embeddings → ② Semantic Similarity → ③ Search with Embeddings → ④ Model Tradeoffs
→ ⑤ Simple Chatbot → ⑥ Temperature & Tokens → ⑦ Prompt Engineering → ⑧ Few-Shot / Zero-Shot
→ ⑨ Simple RAG → ⑩ LangChain Basics → ⑪ What is an Agent?
→ 🎓 Interactive Demos (Streamlit) → 🚀 AgenticAI Foundry

💡 Notebooks marked ✅ No API Key run entirely in Google Colab for free. Notebooks marked 🔑 API Key require an OpenAI key.

---

📚 Tutorials and Notebooks

Welcome to the GenAI Foundry learning series! All notebooks can be opened directly in Google Colab — no local setup required.

Part 1: LLM Mechanics (No API Key Required)

#	📓 Notebook	📝 Description	🔑 API Key	🚀 Colab
1	Tokens and Embeddings	How LLMs break text into tokens and represent meaning as vectors	✅ Free
2	Semantic Similarity	How embeddings capture meaning and measure similarity between text	✅ Free
3	Search with Embeddings	Keyword vs. semantic search, chunking, and the foundations of RAG	✅ Free
4	Model Selection & Tradeoffs	Capability, speed, cost, and context — how to choose the right model	✅ Free

Part 2: Introductory Concepts (OpenAI API Key Required)

#	📓 Notebook	📝 Description	🔑 API Key	⏱️ Time	🚀 Colab
5	Simple Chatbot	Build your first LLM-powered chatbot	🔑 Required	~15 min
6	Temperature & Token Explorer	See how temperature and max tokens change LLM output	🔑 Required	~20 min
7	Prompt Engineering Basics	Zero-shot prompting, system messages, and output formatting	🔑 Required	~25 min
8	Few-Shot / Zero-Shot Prompting	Teaching the model with examples vs. instructions alone	🔑 Required	~20 min
9	Simple RAG Application	Retrieval-Augmented Generation end-to-end	🔑 Required	~30 min
10	LangChain Basics	Chains, prompts, and memory with LangChain	🔑 Required	~25 min
11	Fine-Tuning Basics	Intro to fine-tuning foundation models	🔑 Required	~30 min

Part 3: Bridge to Agentic AI

#	📓 Notebook	📝 Description	🔑 API Key	⏱️ Time	🚀 Colab
12	What is an Agent?	From chatbots to agents — the Observe→Think→Act loop explained	✅ Free	~20 min

---

🎓 Applied Generative AI — Interactive Demos

These demos let you see the concepts from the notebooks in action — no coding required. They are a multi-page Streamlit application running in a single Docker container.

📌 Notebooks → Demos connection: The RAG Visual Explorer (below) visualises exactly what happens in notebooks 3 and 9. The RAG Chat demo is the production version of what you build in notebook 9.

🔍 RAG Visual Explorer — No API key required

Connects to: Notebooks 1, 2, 3, 9

See how Retrieval-Augmented Generation works under the hood:

• Chunking — Watch documents split into searchable pieces
• Embeddings — See text transform into numerical vectors
• Vector Space — Visualise how similar content clusters together
• Semantic Search — Compare keyword vs. meaning-based search

🧠 Ontology & Counterfactual Reasoning — OpenAI API key required for AI features

Connects to: Notebooks 7, 8 (prompt engineering concepts)

Explore structured knowledge and "what if" analysis:

• Ontology Explorer — Interactive knowledge graph of cybersecurity risks
• Counterfactual Analysis — Test what happens when controls fail
• LLM Validation — Verify AI outputs against ground truth

🤖 RAG Chat with Estel — OpenAI API key required

Connects to: Notebooks 9, 10 (RAG + LangChain)

A complete document-grounded chat assistant powered by RAG:

• Upload — Load your own PDF or TXT documents
• Index — Documents are chunked, embedded, and stored in a FAISS vector database
• Chat — Ask questions and get answers grounded in your uploaded content
• Learn — See RAG in action as a complete end-to-end pipeline

📖 How RAG Chat Works — Deep Dive Guide

---

🐳 Running the Demos (Docker Required)

📖 Detailed Docker Guide for Beginners

Step	Command
1. Clone this repo	git clone https://github.com/dlwhyte/GenAI_foundry.git
2. Navigate to repo	cd GenAI_foundry
3. Build container	docker build -t genai-foundry .
4. Run demos	docker run -p 8501:8501 genai-foundry
5. Open browser	http://localhost:8501

With an OpenAI API key (required for Ontology demo + RAG Chat):

docker run -p 8501:8501 -e OPENAI_API_KEY=sk-your-key-here genai-foundry

💡 Without the -e flag, you can still enter your API key in the sidebar on any page that requires it.

🔑 How to Get an OpenAI API Key

💡 Don't have Docker? Download it from docker.com/products/docker-desktop

---

🛠 How to Get Started

New to all of this? Follow these 3 steps:

1. Start with a free notebook — Click any Colab badge in the Part 1 table above. No account or setup needed.
2. Get an API key when ready — Create an OpenAI key to unlock the Part 2 notebooks. Costs pennies per session.
3. Run the interactive demos — Install Docker Desktop, then follow the 5-step table above.

Action	Description
📥 New here?	Download the Student Quick Start Guide (PDF)
✅ New to GitHub?	Start here
✅ Docker setup	Learn how to install and use Docker for running demos
✅ Open notebooks	Open and run the notebooks directly in Google Colab — click any badge above
✅ Get an API key	Create your OpenAI API key to enable model access

---

📝 License

This project is licensed under the MIT License — free to use, modify, and share! See the LICENSE file for more details.

---

🚀 Ready for the Next Level?

This repo covers the foundations of Generative AI — how LLMs work, how to prompt them, and how to build RAG pipelines.

When you're ready to build AI systems that take action in the world, continue to the companion course:

👉 AgenticAI Foundry — Multi-agent systems, tool use, MCP, and agent security

GenAI Foundry covers...	AgenticAI Foundry covers...
Tokens, embeddings, and how LLMs work	How agents observe, think, and act
Prompt engineering and few-shot learning	Multi-agent orchestration with CrewAI
RAG pipelines and vector search	Connecting agents to real tools via MCP
LangChain basics	Agent security and prompt injection defense
Recommended first ✅	Take after GenAI Foundry ➡️