AI Engineering Journey

Quick Links

• 📋 Master Plan - Complete 8-week curriculum
• 📊 Progress Tracker - Track your journey
• 📚 Resources - Curated learning materials

Journey Overview

Week 1: Foundations & Semantic Search          [████░░░░] 45%
Week 2: Vector Databases Deep Dive              [________] 0%
Week 3: Production Vector Search                [________] 0%
Week 4: RAG Systems - Foundations               [________] 0%
Week 5: Advanced RAG Patterns                   [________] 0%
Week 6: MCP Fundamentals                        [________] 0%
Week 7: Autonomous Agents                       [________] 0%
Week 8: Integration & Portfolio                 [________] 0%

Projects You'll Build

1. Semantic Code Search - Week 1
2. Personal Knowledge Base - Week 2
3. Multi-Tenant Search SaaS - Week 3
4. Technical Documentation Assistant - Week 4
5. Adaptive RAG System - Week 5
6. MCP Server Suite - Week 6
7. Research Agent - Week 7
8. Personal AI Platform - Week 8

Folder Structure

ai-engineering-journey/
├── PLAN.md                    # Master plan
├── PROGRESS.md                # Progress tracking
├── README.md                  # This file
├── week-01/                   # Week 1: Semantic Search
│   ├── project/              # Code and deliverables
│   ├── notes/                # Daily logs and learnings
│   └── resources/            # Week-specific resources
├── week-02/                   # Week 2: Vector Databases
├── week-03/                   # Week 3: Production Patterns
├── week-04/                   # Week 4: RAG Foundations
├── week-05/                   # Week 5: Advanced RAG
├── week-06/                   # Week 6: MCP Servers
├── week-07/                   # Week 7: Autonomous Agents
├── week-08/                   # Week 8: Integration
├── portfolio/                 # Portfolio artifacts
│   ├── blog-posts/           # Weekly blog posts
│   ├── videos/               # Demo videos
│   └── demos/                # Live demos
└── resources/                 # Global resources
    ├── reading/              # Reading materials
    ├── tools/                # Tool configurations
    └── templates/            # Code templates

Current Week: Week 1 - In Progress (45% Complete)

✅ Completed (Days 1-4)

Learning Phase (30+ hours)

• Day 1: Embeddings fundamentals - RNNs, LSTMs, Transformers (5+ hours)
• Day 2-3: Deep dive into LLM architecture, training, and alignment (8+ hours)
• Day 3: Finalized mental models and learning documentation

Planning Phase (In Progress)

• Day 4: Created Experiment 1 with 7 validation tests
  • SPEC.md and README.md with clear methodology
  • 7 test files (pre-training, semantic clustering, dimensionality, distance metrics, relationships, chunking, working memory)
  • day4readingnotes.md with detailed learning objectives for each test
  • Infrastructure: run_all.py, results.md template, requirements.txt

🔄 In Progress (Day 5-7)

• Day 5: Run 7 validation tests and document findings
• Day 6: Build rag-code-qa project with validated architecture
• Day 7: Finalize, polish, and ship

Learning Documentation

• day1-reading-notes.md - Embeddings fundamentals
• Date2-3-Deep-Dive-Notes.md - LLM architecture & training
• day4readingnotes.md - Experiment learning objectives
• daily-log.md - Day-by-day progress

Architecture Decisions (Informed by Experiments)

• Embedding Model: text-embedding-3-small (1536 dimensions)
• Distance Metric: Cosine similarity (optimal for embeddings)
• Chunking Strategy: By function (semantic units > fixed-size)
• Multi-language Support: Python, JavaScript, TypeScript
• Caching: Safe and necessary (embeddings are pre-computed)

Next: Run Experiments

cd week-01/project/experiments/01_embeddings
python run_all.py

Tech Stack

Languages

• Python - AI/ML, data processing
• TypeScript - MCP servers, tooling

Vector Databases

• ChromaDB (Week 1)
• Qdrant (Week 2)
• Pinecone (Week 3)
• pgvector (Week 2)

LLMs & APIs

• Claude API (primary)
• OpenAI (embeddings, GPT-4)
• Cohere (reranking)

Frameworks

• Click (CLI)
• FastAPI (APIs)
• Next.js (Frontend)
• LangChain (comparative study)

Daily Workflow

1. Morning: Read and plan (30-60 min)
2. Code: Build and ship (2-4 hours)
3. Document: Write notes (15-30 min)
4. Share: Post on LinkedIn/Twitter (5 min)

Content Strategy

• Daily: Social media updates
• Weekly: Technical blog post
• Bi-weekly: Video demo
• End: Portfolio website

Success Criteria

• ✅ 8 projects shipped and deployed
• ✅ 8 blog posts published
• ✅ Portfolio website live
• ✅ Deep understanding of AI engineering

Let's Build! 🚀

Start with PLAN.md and begin your journey.