A 15-week curriculum to become a job-ready AI Engineer.
No prior AI/ML experience required. You start from your first API call and end with a production-ready application and 3 portfolio projects.
Week 1: "I can call LLM APIs, build a chatbot, and get structured output"
Weeks 2-4: "My AI answers questions from my own documents (RAG)"
Weeks 5-6: "I can measure quality and protect my system from attacks"
Weeks 7-11: "My AI can reason, take actions, and use tools autonomously"
Weeks 12-15: "My AI system is production-ready and I'm interview-ready"
See agenda/00_summary.md for the complete roadmap with priorities and interview relevance.
| Week | Topic | What You'll Build |
|---|---|---|
| Week 1 | API Basics, Chatbot & Structured Output | First API call, chatbot with memory, validated JSON output |
| Week 2 | RAG Introduction | Document Q&A — your chatbot answers from your own files |
| Week 3 | Embeddings Deep Dive | Semantic search pipeline, embedding model comparison |
| Week 4 | Production RAG Architecture | Full RAG with Qdrant, citations, hybrid search |
| Week | Topic | What You'll Build |
|---|---|---|
| Week 5 | Evaluations & Quality Metrics | Test suite, retrieval/generation metrics, LLM-as-judge |
| Week 6 | Security & Guardrails | Input validation, PII protection, red-team tests — Portfolio #1 |
Portfolio #1: Secure HR Q&A Bot — RAG + security guardrails + evaluations
| Week | Topic | What You'll Build |
|---|---|---|
| Week 7 | Agent Fundamentals (ReAct) | Custom ReAct agent, multi-step problem solver |
| Week 8 | Tool Calling & Function Execution | AI that calls multiple tools, full tool loop |
| Week 9 | Agent Frameworks — LangGraph & Multi-Modal | LangGraph workflows, Strands Agents multi-agent, vision/audio (optional) |
| Week 10 | LangGraph in Practice + LangSmith & Langfuse | Observable, evaluatable agents with tracing |
| Week 11 | Data Engineering for AI | SQL, ETL pipelines, data quality — Portfolio #2 |
Portfolio #2: AI Agent — ReAct reasoning + tools + step logging + data pipeline
| Week | Topic | What You'll Build |
|---|---|---|
| Week 12 | MCP & Observability | MCP server, logging, tracing, metrics dashboard |
| Week 13 | Production Hardening | Rate limiting, caching, retries, RBAC, secrets management |
| Week 14 | Deployment & Fine-Tuning | Docker, cloud deployment, Slack/Teams bot |
| Week 15 | Capstone & Interview Prep | Final project + interview readiness — Portfolio #3 |
Portfolio #3: Full-Stack AI Research Assistant — FastAPI + RAG + agents + Docker
# Install dependencies (uses uv)
uv sync
# Copy and fill in your API keys
cp .env.example .env
# Add .env keys
# Start Jupyter Lab
uv run jupyter labOPENAI_API_KEY=sk-...
# Optional — use local Ollama models instead of OpenAI
AI_PROVIDER=ollama # "openai" (default) or "ollama"
OLLAMA_BASE_URL=http://localhost:11434
All weeks support Ollama as a free alternative to OpenAI:
# Install Ollama, then pull models as needed
ollama pull llama3.1:8b # General chat (CHAT_MODEL)
ollama pull nomic-embed-text # Embeddings (EMBED_MODEL)
ollama pull qwen2.5-coder:7b # Code generation (CODE_MODEL)
ollama pull llama3.2-vision:11b # Vision/multimodal (VISION_MODEL)
ollama pull deepseek-r1:8b # Reasoning/ReAct agents (REASON_MODEL)
ollama pull qwen2.5:7b # Structured output (STRUCTURED_MODEL)
ollama pull granite3-guardian:2b # Safety/guardrails (SAFETY_MODEL)
# Switch provider
export AI_PROVIDER=ollamaThe unified client in scripts/model_config.py handles provider switching automatically.
Each week has a companion Jupyter notebook in labs/:
| Lab | Topic |
|---|---|
| week_00_getting_started.ipynb | Environment setup verification |
| week_01_first_api_call.ipynb | API basics, chatbot, structured output |
| week_02_rag_intro.ipynb | RAG pipeline introduction |
| week_03_embeddings.ipynb | Embeddings and semantic search |
| week_04_production_rag.ipynb | Production RAG with Qdrant |
| week_05_evaluations.ipynb | Evaluations and quality metrics |
| week_06_security.ipynb | Security and guardrails |
| week_07_agents.ipynb | Agent fundamentals (ReAct) |
| week_08_tool_calling.ipynb | Tool calling and function execution |
| week_09_frameworks.ipynb | LangGraph and agent frameworks |
| week_10_frameworks_observability.ipynb | LangSmith and Langfuse |
| week_11_data_engineering.ipynb | SQL, ETL pipelines, data quality |
| week_12_mcp.ipynb | MCP and observability |
| week_13_production.ipynb | Production hardening |
| week_14_deployment.ipynb | Deployment and fine-tuning |
| week_15_capstone.ipynb | Capstone and interview prep |
- Read the agenda first:
agenda/week_XX.md - Open the lab notebook:
labs/week_XX_*.ipynb - Try each exercise yourself — hints and solutions are collapsible inside the notebook
| # | Project | Week | Skills Demonstrated |
|---|---|---|---|
| 1 | Secure HR Q&A Bot | 6 | RAG, embeddings, security, evaluations |
| 2 | AI Agent | 11 | Tool calling, ReAct, observability, data pipelines |
| 3 | Full-Stack AI Research Assistant | 15 | All of the above + FastAPI + Docker |
ai_track/
├── agenda/ # Week-by-week guides with theory and tasks
│ ├── 00_summary.md # Full curriculum overview and roadmap
│ └── week_01.md ... week_15.md
├── labs/ # Jupyter notebook exercises (one per week)
│ └── week_00_getting_started.ipynb ... week_15_capstone.ipynb
├── projects/ # Portfolio project templates
│ ├── 01_rag_qa_system/ # Portfolio #1: Secure HR Q&A Bot
│ ├── 02_ai_agent/ # Portfolio #2: AI Agent
│ └── 03_capstone/ # Portfolio #3: Full-Stack Research Assistant
├── scripts/
│ └── model_config.py # Unified OpenAI/Ollama client
├── knowledges.md # Quick-reference cheat sheet for all concepts
├── pyproject.toml # Dependencies (managed with uv)
└── uv.lock
- Basic Python (variables, functions, loops, classes)
- Command line basics
- No AI/ML experience needed
- agenda/00_summary.md — Full roadmap with priorities and interview relevance
- knowledges.md — Concept quick-reference cheat sheet
- Each week's agenda includes links to relevant docs and tutorials