100 Days · 5 Phases · 50+ Hands-on Projects · 300+ Free Resources
| Phase | Days | Main Focus |
|---|---|---|
| 🟣 Phase 1 | 1–20 | Foundations: Python, LLMs, APIs, First Chatbot |
| 🔵 Phase 2 | 21–40 | RAG, Image, Voice, LangChain, FastAPI |
| 🟢 Phase 3 | 41–60 | AI Agents, CrewAI, LangGraph, Automation |
| 🟡 Phase 4 | 61–80 | Fine-tuning, Docker, Cloud Deploy, SaaS |
| 🔴 Phase 5 | 81–100 | Product Launch, Marketing, Scaling |
Learn Python, LLMs, and APIs — the cornerstone of everything that follows
Understand the difference between AI, ML, DL, and GenAI. Watch a comprehensive intro before writing any code.
Beginner
If you're not fluent in Python, learn everything you need in one day — Lists, Dicts, Functions, Classes.
Python
Set up Anaconda or use Google Colab. Learn how to write and experiment with code quickly.
Setup
Understand Transformers, Tokenization, and Attention. Andrej Karpathy's most famous video ever.
LLM Theory
Sign up for OpenAI, get your API Key, and build your first Python script that talks to GPT.
API
Learn Zero-shot, Few-shot, and Chain of Thought prompting. A great prompt is worth an entire codebase.
Prompting
- 📚 DeepLearning.AI: Prompt Engineering for Developers (Free)
- 🌐 Prompt Engineering Guide
- 🎬 Prompt Engineering Full Guide
Try the Claude API and compare it with GPT. Learn the Anthropic SDK and System Prompts.
Claude
Understand how tokens are counted and how to choose the right model. Save API costs intelligently.
API · Costs
Build a Terminal Chatbot that maintains Conversation History. Your first real application.
Project · Python
- 🎬 Build Chatbot Python + OpenAI
- 📚 DeepLearning.AI: Building Systems with ChatGPT API (Free)
- 🎬 Python Chatbot Tutorial
Explore the Hugging Face Hub. Try free models for Text, Image, and Audio tasks.
HuggingFace
Understand Vector Embeddings — how to convert text to numbers and use them for semantic search.
Embeddings
- 🎬 Word Embeddings Explained
- 🎬 OpenAI Embeddings Tutorial
- 📚 DeepLearning.AI: Vector Databases & Embeddings (Free)
Store embeddings in a database and search them by meaning, not keyword. The foundation of RAG.
VectorDB
Turn your code into a beautiful Web App in minutes using Streamlit — no HTML required.
UI · Project
Combine what you've learned: Streamlit + OpenAI API. Build a Text Summarizer or Chatbot.
Project · Full App
Learn LangChain from scratch: Chains, PromptTemplates, OutputParsers.
LangChain
Add memory to your Chatbot: ConversationBufferMemory, SummaryMemory, and the difference between them.
LangChain · Memory
Make LLMs return structured JSON. Use Pydantic with LangChain to extract clean data.
Structured
- 📚 DeepLearning.AI: Structured Outputs with Pydantic (Free)
- 🎬 JSON Output from LLM
- 📚 LangChain Output Parsers Docs
Save your code on GitHub. Learn branches, requirements.txt, and env files.
Tools
Run LLMs on your machine for free using Ollama: Llama 3, Mistral, Phi.
Open Source · Local
Build a complete app: Simple PDF Chatbot using everything you've learned.
Milestone · Project
Build Chatbots, RAG Systems, and real applications using the most popular frameworks
Understand RAG from scratch: how to make an LLM answer accurately from your own data.
RAG
Document Loaders + Text Splitters + Embeddings + Vector Store + Retriever. Code from scratch.
RAG · Project
Learn how to prepare any file type for RAG: PDFs, Websites, Excel, Notion.
RAG · Data
Bad chunking ruins RAG. Learn Semantic Chunking and Recursive Splitting.
RAG · Advanced
Move from local Chroma to cloud-hosted Pinecone. Build a RAG that runs on the internet.
VectorDB · Cloud
Build a professional demo interface for any AI model in 5 minutes with Gradio.
UI
- 📚 DeepLearning.AI: Building GenAI Apps with Gradio (Free)
- 📚 Gradio Documentation
- 🎬 Gradio Complete Tutorial
Use the DALL-E 3 API and build an Image Generator. Explore Stable Diffusion for free.
Image AI
- 🎬 DALL-E API Tutorial
- 📚 DeepLearning.AI: How Diffusion Models Work (Free)
- 🌐 Stability AI Platform Docs
Convert any audio to text using OpenAI's Whisper. Build a Voice Transcriber.
Audio
Convert text to natural human-sounding voice using ElevenLabs or OpenAI TTS.
Audio
Combine Whisper + LLM + ElevenLabs in one app: speak to AI and it responds with voice.
Project · Voice
Make LLMs call external APIs and execute code. The foundation for building AI Agents.
Function Calling
Learn LlamaIndex as a powerful alternative to LangChain for building RAG systems.
LlamaIndex
Build a REST API for your model using FastAPI. Let any application consume your AI.
FastAPI · Backend
Use GPT-4V or Claude Vision to analyze images, describe them, and answer questions about them.
Vision · Multimodal
Project: Upload images or PDFs and query them — Financial Reports Analyzer or Invoice Extractor.
Project · Vision
Learn BLEU, ROUGE, and LLM-as-Judge. The cure for hallucinations.
Evaluation
Understand the risks of Prompt Injection and how to protect your app from attacks and jailbreaks.
Security
Save 90% of API costs with Semantic Caching. Learn GPTCache and optimization techniques.
Performance
Track and monitor your AI app's performance using LangSmith or Weights & Biases.
MLOps
Build a complete Customer Support Bot: RAG + Memory + FastAPI + Streamlit UI.
Milestone · Full Stack
Build intelligent Agents that act autonomously. Learn LangGraph, CrewAI, and automation tools
What's the difference between a Chain and an Agent? Learn the ReAct Pattern and how agents plan and execute.
Agents
Build an Agent using OpenAI + Function Calling — no framework needed.
Agents · Project
Learn LangGraph to build complex agents with State Management and Conditions.
LangGraph
- 📚 DeepLearning.AI: AI Agents in LangGraph (Free)
- 📚 LangGraph Official Docs
- 🎬 LangGraph Complete Tutorial
Give your agent tools: Web Search, Calculator, Database, Email, Calendar.
Agents · Tools
Build a team of Agents, each with a different role, that collaborate together.
CrewAI · Multi-Agent
- 📚 CrewAI Official Docs
- 📚 DeepLearning.AI: Multi-AI Agents with CrewAI (Free)
- 🎬 CrewAI Complete Tutorial
Build an agent that searches the internet, collects data, and writes reports automatically.
Agents · Project
Build an agent that writes Python, runs it, and fixes its own errors — like ChatGPT's Code Interpreter.
Agents · Code
Build intelligent automation workflows without code using n8n: Email, Slack, Database, AI.
Automation · No-Code
Use the Assistants API to build powerful agents with File Search and Code Interpreter.
Assistants · OpenAI
Review everything you've learned, organize your GitHub profile, and build a portfolio project.
Milestone
Try AutoGen to build agents that talk to each other and solve complex problems.
AutoGen · Multi-Agent
Build an agent that opens websites, fills forms, and clicks buttons — AI-powered automation.
Browser · Automation
Explore Claude Computer Use and learn how AI can see the screen and interact with it.
Computer Use
Learn the new protocol that lets AI models communicate with any tool or service.
MCP
Build an agent that decides when to search documents and when to use its general knowledge.
RAG · Agents
- 📚 DeepLearning.AI: Agentic RAG with LlamaIndex (Free)
- 🎬 Agentic RAG Tutorial
- 🎬 Self-RAG Paper Explained
Agents make decisions: Router Chains, Conditional Branching, Self-Reflection.
LangGraph · Advanced
Write Unit Tests for prompts and Evaluations for RAG. Quality assurance for AI apps.
Testing
Protect your app from harmful content using Guardrails AI and NeMo Guardrails.
Safety
Make your app display answers as they're being typed — like ChatGPT. Server-Sent Events and WebSockets.
Streaming
Build a Research Agent: searches the internet, reads articles, and writes a comprehensive report.
Milestone · Full Agent
Fine-tune models, build APIs, deploy to the cloud, and learn MLOps for production
When should you fine-tune vs. use Prompt Engineering?
Fine-tuning
How to prepare your data: JSONL format, Data Cleaning, Train/Val Split.
Fine-tuning · Data
Execute a real fine-tuning job on GPT-3.5 Turbo. Step-by-step walkthrough.
Fine-tuning · OpenAI
Fine-tune large models on a small GPU using LoRA and QLoRA. Free on Google Colab!
LoRA · Open Source
- 🎬 LoRA Fine-tuning Tutorial
- 📚 Unsloth: Fast Fine-tuning (GitHub)
- 📚 DeepLearning.AI: Efficiently Serving LLMs (Free)
Use the PEFT library for Parameter-Efficient Fine-Tuning on any model.
HuggingFace · Fine-tuning
Put your app in a Docker container. Learn Dockerfile, docker-compose, and ENV vars.
Docker · DevOps
Ship your app to the cloud. Learn EC2, Lambda, and Cloud Run — free tier to start.
Cloud · Deploy
The easiest way to deploy: Vercel, Railway, and Render. Ship in two minutes.
Deploy · Serverless
Build an automated pipeline: GitHub Actions → Test → Build → Deploy on every push.
DevOps · CI/CD
Store User Data, Conversations, and Embeddings in a real database: PostgreSQL and MongoDB.
Database
Add a login system: Auth0, Supabase Auth, or Firebase.
Auth
Protect your API from abuse: Rate Limiting, API Keys, Usage Tracking.
API · Security
Monitor your app: Errors, Performance, Cost. Use Sentry and Prometheus.
Monitoring
Cut AI API costs: Batching, Caching, Model Selection, Compression.
Costs · Optimization
Handle thousands of users: Load Balancing, Queue Systems, Horizontal Scaling.
Scale · Advanced
Build an AI app with Next.js and React. The Vercel AI SDK makes streaming trivially easy.
Frontend · Next.js
Bring your AI app to mobile using React Native + Expo + AI APIs.
Mobile
Track experiments and results with MLflow. Essential for every fine-tuning project.
MLOps
Compare different models in production and choose the best one based on real data.
Testing · Production
Build a small SaaS Tool ready to sell: Fine-tuned Model + Auth + Billing + Deploy.
Milestone · SaaS
Bring everything together to build a complete AI product from idea to launch
How do you choose a winning AI product idea? Learn Product-Market Fit and User Research.
Product
Study competitors, understand differentiation, and identify the niche you'll own.
Strategy
Talk to 10 users before writing a single line of code. Validation saves months of wasted work.
Research
Build a real MVP in a week. Learn Rapid Prototyping and the art of shipping fast.
MVP · Ship
Build a landing page that converts visitors to users. Learn how to write AI marketing copy.
Marketing
Freemium, Subscription, Pay-per-Use. Stripe Integration and Pricing Strategy.
Business · Revenue
Get people to find your tool on Google. Learn Technical SEO and Content Strategy.
Marketing · SEO
Build a waitlist, launch on ProductHunt, and work Reddit and Twitter/X.
Launch
Collect feedback, analyze data, and make decisions based on data — not opinions.
Product
Learn how to handle growth: Infrastructure Scaling and Onboarding Automation.
Scale · Growth
Bias Detection, Fairness, Transparency. How to build ethical and responsible AI.
Ethics
Understand data protection laws and how to handle user data legally and responsibly.
Legal
Meta-Prompting, Self-Consistency, Tree of Thoughts, Reflexion — for professionals.
Advanced · Prompting
- 🌐 Prompt Engineering Guide: Advanced Techniques
- 🎬 Advanced Prompting 2025
- 📚 DeepLearning.AI: Prompt Engineering with Llama 2 (Free)
Learn how to build and train your own model from zero. A GPT-like model with 20M parameters.
Advanced · Research
- 🎬 Karpathy: Let's Build GPT from Scratch
- 📚 DeepLearning.AI: Pretraining LLMs (Free)
- 🎬 Train an LLM Step by Step
Explore AI applications across different industries and the major opportunities they offer.
Industry · Vertical AI
AI changes every day. Learn how to follow the latest news and research consistently.
Learning
Work openly on X/Twitter and LinkedIn. This attracts customers and investors alike.
Brand
Contribute to open-source AI projects: LangChain, Hugging Face, Ollama.
Open Source
Organize GitHub, build a personal website, and write case studies for your best projects.
Career
Launch your complete project on ProductHunt and share it with the world. You are now a professional AI Builder!
🎉 Launch Day
| Channel | Specialization | Link |
|---|---|---|
| Andrej Karpathy | LLM Theory, Build from Scratch | YouTube |
| 3Blue1Brown | Math & Neural Networks (Animations) | YouTube |
| IBM Technology | AI Concepts & Explainers | YouTube |
| AssemblyAI | RAG, Voice AI, LLM Engineering | YouTube |
| LangChain | LangChain, LangGraph, Agents | YouTube |
| DeepLearning.AI | Courses & Tutorials | YouTube |
| sentdex | Python & AI Projects | YouTube |
| Matt Wolfe | AI News & Tools | YouTube |
| Sam Witteveen | LLMs & Agents | YouTube |
| Nicholas Renotte | AI Apps & Projects | YouTube |
| Resource | Description | Link |
|---|---|---|
| DeepLearning.AI | Best free AI courses (Andrew Ng) | deeplearning.ai |
| Hugging Face | Open-source models & free courses | huggingface.co |
| Prompt Engineering Guide | Comprehensive prompting reference | promptingguide.ai |
| LangChain Docs | Official LangChain documentation | python.langchain.com |
| OpenAI Docs | Official OpenAI API documentation | platform.openai.com |
| Anthropic Docs | Official Claude API documentation | docs.anthropic.com |
| Kaggle | Competitions & free courses | kaggle.com |
| Pinecone Learn | Learn Vector Databases | pinecone.io/learn |
| ArXiv | Latest AI research papers | arxiv.org |
| The Batch | Weekly AI newsletter | deeplearning.ai/the-batch |
| Tool | Use Case |
|---|---|
| LangChain | Building LLM Applications |
| LlamaIndex | RAG Systems |
| LangGraph | Stateful AI Agents |
| CrewAI | Multi-Agent Systems |
| Streamlit | Rapid Web UI |
| FastAPI | REST APIs |
| Pinecone / Chroma | Vector Databases |
| Ollama | Local LLMs |
| n8n | No-Code Automation |
| Docker | Containerization |
| Vercel AI SDK | Full Stack AI Web Apps |
🚀 Made with ❤️ — All links are free and open-source | 2025–2026