🧭 Roadmap
Home / 00 · Kickstart / Your Complete AI Learning Path
🚀 00 · Kickstart

🎯 Your Complete AI Learning Path

Overview: From Zero to AI Expert

This guide shows you exactly how to progress from Kickstart through Foundations to master AI concepts.

📚 Stage 0: Kickstart (10-15 hours)

Goal: Understand WHY AI works now and get historical context

Week 1: Foundation & Context (7 hours)

Day 1-2: Historical Context (2 hours)

Day 3-4: Understanding Scale (3 hours)

Day 5-6: The Breakthroughs (4 hours)

  • 📖 03-The-Three-Breakthroughs.md ⭐
  • Learn: Scale, Training, Architecture
  • Learn: Why it happened in 2020
  • Learn: Hardware + Data + Algorithms convergence

Week 2: Practical Understanding (5.5 hours)

Day 7: Paradigm Shift (1.5 hours)

  • 📖 02-Why-AI-Now.md
  • Learn: Traditional vs AI approach
  • Learn: What LLMs solved
  • Learn: Real-world impact

Day 8-9: Architecture (2 hours)

Day 10: Evolution (2 hours)

Week 3: Impact & Future (5 hours)

Day 11-12: Real Impact (2 hours)

Day 13: Future Vision (1.5 hours)

  • 📖 08-What-Comes-Next.md
  • Learn: 2024-2030 predictions
  • Learn: Emerging trends
  • Learn: Career opportunities

Day 14: Hands-On Project (3 hours)

🎓 Stage 1: Foundations (25-30 hours)

Goal: Master the technical fundamentals of how AI actually works

Week 4-5: Core Concepts (15-20 hours)

Topic 1: Tokenization (3-4 hours)

  • 📖 01-Foundations/01-Tokenization.md
  • Learn: How text becomes numbers
  • Learn: BPE, WordPiece, SentencePiece
  • Build: Your own tokenizer
  • Why this matters: Everything starts with tokenization

Topic 2: Vectorization (3-4 hours)

  • 📖 01-Foundations/02-Vectorization/_index.md
  • Learn: Word embeddings
  • Learn: Semantic similarity
  • Learn: Vector spaces
  • Why this matters: How AI understands meaning

Topic 3: Attention Mechanism (4-5 hours)

  • 📖 01-Foundations/03-Attention.md
  • Learn: Self-attention
  • Learn: Multi-head attention
  • Learn: Why attention works
  • Why this matters: The breakthrough that changed everything

Topic 4: Transformer Architecture (5-6 hours)

  • 📖 01-Foundations/04-Transformer.md
  • Learn: Complete architecture
  • Learn: Encoder-Decoder
  • Learn: Position encoding
  • Why this matters: Foundation of all modern LLMs

Topic 5: Self-Supervised Learning (3-4 hours)

  • 📖 01-Foundations/05-Self-Supervised-Learning.md
  • Learn: Pre-training methods
  • Learn: Masked language modeling
  • Learn: Next token prediction
  • Why this matters: How LLMs learn from unlabeled data

Topic 6: Large Language Models (3-4 hours)

  • 📖 01-Foundations/06-Large-Language-Models.md
  • Learn: LLM architecture
  • Learn: Training process
  • Learn: Inference
  • Why this matters: Putting it all together

🚀 Stage 2: Intermediate (20-25 hours)

Goal: Learn advanced techniques for using and improving LLMs

Week 6-7: Advanced Techniques

Topic 7: Fine-Tuning (3-4 hours)

  • Learn: Adapting pre-trained models
  • Learn: LoRA, QLoRA
  • Practice: Fine-tune a model

Topic 8: Few-Shot Prompting (2-3 hours)

  • Learn: In-context learning
  • Learn: Prompt engineering
  • Practice: Effective prompting

Topic 9: Context Engineering (2-3 hours)

  • Learn: Managing context windows
  • Learn: Context optimization
  • Practice: Long-form tasks

Topic 10: Chain-of-Thought (2-3 hours)

  • Learn: Reasoning techniques
  • Learn: Step-by-step problem solving
  • Practice: Complex reasoning

Topic 11: Vector Databases (3-4 hours)

  • Learn: Semantic search
  • Learn: Embeddings storage
  • Practice: Build vector DB

Topic 12: RAG (Retrieval-Augmented Generation) (4-5 hours)

  • Learn: Combining retrieval + generation
  • Learn: RAG architecture
  • Practice: Build RAG system

🎯 Stage 3: Advanced (25-30 hours)

Goal: Master cutting-edge AI techniques

Week 8-10: Expert Topics

Topic 13: AI Agents (4-5 hours)

  • Learn: Autonomous agents
  • Learn: Tool use
  • Practice: Build an agent

Topic 14: Reinforcement Learning (4-5 hours)

  • Learn: RLHF
  • Learn: Reward models
  • Practice: Align a model

Topic 15: Reasoning Models (3-4 hours)

  • Learn: Chain-of-thought
  • Learn: Tree-of-thought
  • Practice: Complex reasoning

Topic 16: Multi-Modal Models (4-5 hours)

  • Learn: Vision + Language
  • Learn: Audio + Text
  • Practice: Multi-modal tasks

Topic 17: Small Language Models (3-4 hours)

  • Learn: Distillation
  • Learn: Efficient models
  • Practice: Deploy small model

Topic 18: Distillation (3-4 hours)

  • Learn: Knowledge transfer
  • Learn: Model compression
  • Practice: Distill a model

Topic 19: Quantization (3-4 hours)

  • Learn: Model optimization
  • Learn: INT8, INT4
  • Practice: Quantize a model

Topic 20: Model Context Protocol (3-4 hours)

  • Learn: MCP architecture
  • Learn: Tool integration
  • Practice: Build MCP server

🌟 Stage 4: Emerging AI (15-20 hours)

Goal: Stay ahead with cutting-edge research

Week 11-12: Future Technologies

Topic 21: Diffusion Models (3-4 hours)

  • Learn: Image generation
  • Learn: Stable Diffusion
  • Practice: Generate images

Topic 22: Vision Transformers (3-4 hours)

  • Learn: ViT architecture
  • Learn: Image understanding
  • Practice: Image classification

Topic 23: Mixture of Experts (3-4 hours)

  • Learn: MoE architecture
  • Learn: Efficient scaling
  • Practice: Understand routing

Topic 24: Constitutional AI (3-4 hours)

  • Learn: AI safety
  • Learn: Alignment techniques
  • Practice: Safe AI design

📊 Complete Timeline

Month 1: Kickstart (Context & History)
├── Week 1: Historical foundation
├── Week 2: Practical understanding
└── Week 3: Impact & hands-on

Month 2: Foundations (Technical Fundamentals)
├── Week 4: Tokenization + Vectorization
├── Week 5: Attention + Transformers
└── Week 6: Self-supervised + LLMs

Month 3: Intermediate (Advanced Techniques)
├── Week 7: Fine-tuning + Prompting
├── Week 8: Context + Chain-of-thought
└── Week 9: Vector DB + RAG

Month 4: Advanced (Expert Topics)
├── Week 10: Agents + RL
├── Week 11: Reasoning + Multi-modal
└── Week 12: SLMs + Optimization

Month 5: Emerging (Cutting Edge)
├── Week 13: Diffusion + Vision
└── Week 14: MoE + Constitutional AI

🎯 Learning Paths by Goal

Path 1: "I want to USE AI" (Fast Track - 6 weeks)

  1. Kickstart (all files) - 2 weeks
  2. Foundations (01-03 only) - 2 weeks
  3. Intermediate (08, 12 only) - 1 week
  4. Build projects - 1 week

Path 2: "I want to UNDERSTAND AI" (Standard - 3 months)

  1. Kickstart (all files) - 3 weeks
  2. Foundations (all topics) - 5 weeks
  3. Intermediate (all topics) - 4 weeks

Path 3: "I want to BUILD AI" (Complete - 5 months)

  1. Kickstart - 3 weeks
  2. Foundations - 5 weeks
  3. Intermediate - 4 weeks
  4. Advanced - 6 weeks
  5. Emerging - 3 weeks

💡 Study Tips

Daily Routine

  • Morning (1 hour): Read theory
  • Afternoon (1 hour): Code examples
  • Evening (30 min): Review & notes

Weekly Goals

  • Complete 2-3 topics
  • Build 1 small project
  • Review previous week

Monthly Milestones

  • Month 1: Understand AI history & context
  • Month 2: Master technical fundamentals
  • Month 3: Apply advanced techniques
  • Month 4: Build expert-level projects
  • Month 5: Explore cutting-edge research

✅ Progress Tracking

Use this checklist:

Kickstart ✓

  • 00-AI-Winter-to-Summer
  • 01-LM-vs-LLM-Deep-Dive
  • 03-Three-Breakthroughs
  • 02-Why-AI-Now
  • 05-Traditional-vs-AI-Servers
  • 06-BERT-to-GPT-Revolution
  • 07-Real-World-Impact
  • 08-What-Comes-Next
  • 09-Your-First-AI-Server

Foundations ✓

  • 01-Tokenization
  • 02-Vectorization
  • 03-Attention
  • 04-Transformer
  • 05-Self-Supervised-Learning
  • 06-Large-Language-Models

Intermediate ✓

  • 07-Fine-Tuning
  • 08-Few-Shot-Prompting
  • 09-Context-Engineering
  • 10-Chain-of-Thought
  • 11-Vector-Database
  • 12-RAG

Advanced ✓

  • 13-Agents
  • 14-Reinforcement-Learning
  • 15-Reasoning-Models
  • 16-Multi-Modal-Models
  • 17-Small-Language-Models
  • 18-Distillation
  • 19-Quantization
  • 20-Model-Context-Protocol

Emerging ✓

  • 21-Diffusion-Models
  • 22-Vision-Transformers
  • 23-Mixture-of-Experts
  • 24-Constitutional-AI

🚀 Ready to Start?

Your first step: Open 00-The-AI-Winter-to-Summer.md

Time commitment: 15-30 minutes per day = Complete in 3-5 months

Remember:

  • Learn at your own pace
  • Build projects along the way
  • Review regularly
  • Ask questions
  • Have fun! 🎉

Last Updated: 2024-03-14 Total Learning Time: 90-120 hours Difficulty: Beginner → Expert