BUILD & LEARN · DATA SCIENCE SATURDAYS
AI Engineering
WEEK 1
Psychology · Setup · Basic Code Structure · AI Tooling (CLI) · OpenAI Key
Saturday · 10:00–12:00 · Berlin
Hosted by Lindsey Peng
About This Course
A seven-week, hands-on program to build and deploy a real AI application — end to end. This is not theory-only machine learning: you'll ship working code touching FastAPI, Postgres, Docker, and a real AI pipeline.
WHO IT'S FOR
WHAT YOU'LL TOUCH
FastAPI
Postgres
Docker
Real AI Pipeline
7 Weeks → 1 Deployed AI App in Your Own GitHub Repo
Saturdays · 10:00–12:00 · Octopus Bar, Berlin
AI Engineering · Week 1
1
Week 1 Roadmap
Psychology
Mindset shift: from practitioner to orchestrator
Setup
Environment: Python, Git, Docker
Basic Code Structure
Folders, placeholders, CLAUDE.md
AI Tooling (CLI)
Claude CLI workflow
OpenAI Key
API access & .env config
AI Engineering · Week 1
2
PART 1
Psychology
AI Engineering · Week 1
3
Why Psychology Comes First
Before we touch a single tool, we need to address the mental model. This course isn't about memorizing new syntax — it's about reshaping how you approach building software in the AI era.
Breaking Through the Fear
From practicing one skill to becoming an orchestrator of many
AI Engineering · Week 1
4
PSYCHOLOGY
Breaking Through the Fear
“I haven't done this before.”
Like learning to ride a bicycle or stand on a skateboard for the first time — the wobble is part of the process, not a sign you're doing it wrong.
This is a new era, and it requires a complete transition: from practicing a specific skill to becoming an orchestrator.
That means understanding, at a high level, how tools work together and how data travels between them — rather than the fine detail of how any single tool works internally.
PRACTITIONER
Knows one tool deeply
ORCHESTRATOR
Understands how the whole system fits together
AI Engineering · Week 1
5
THE APPROACH
Training You Two Ways
Becoming an AI engineer takes both reps and a real deliverable. This course trains you on two parallel tracks:
PRACTICING
Work WITH AI
Using AI in your daily workflow — including tools like Claude Code — until it becomes second nature.
BUILDING
An AI-Powered Service
Together, we build a real service where AI isn't bolted on — it's a core part of the workflow.
AI Engineering · Week 1
6
THE APPROACH · PRACTICING
Work WITH AI
This isn't just a project to use AI as your main service — it's about working WITH AI in your daily workflow until it becomes second nature. That's exactly why we highly recommend Claude Code.
We start from nothing but an idea in our head — and ask AI to structure everything from there.
NOT THIS
AI as a tool you occasionally call for a task
INSTEAD
AI as your structuring partner, from first idea to working implementation
AI Engineering · Week 1
7
THE APPROACH · PRACTICING · BONUS
Optional but Highly Recommended
OPTIONAL · HIGHLY RECOMMENDED
Claude Pro — $20/month
The Claude desktop app for writing, research, and problem-solving — plus Claude Code, a terminal-based AI coding agent that lets you delegate entire development tasks straight from your command line.
Claude Max — $100 or $200/month
For heavy users: 5–20x more usage than Pro, plus access to the most powerful Claude models.
This single subscription lets you ship faster, handle more complex projects, and operate like a small team. The ROI is a no-brainer.
AI Engineering · Week 1
8
THE APPROACH · BUILDING
An AI-Powered Service
This isn't a toy exercise — together, we build a real service where AI is a core part of the workflow, not a bolt-on feature.
By the end of this course, you'll have shipped an AI News API: scraping, enrichment, storage, and an API — all built together, step by step.
NOT THIS
A disposable tutorial project
INSTEAD
A real, working service — live in your own GitHub repo
AI Engineering · Week 1
9
PART 2
Actual Project Setup
AI Engineering · Week 1
10
BEFORE YOU START
Two Paths to Follow Along
Build From Scratch
Follow every step this week: install the tools yourself, scaffold the project by hand, and use Claude to help structure it from a blank idea.
USING CLAUDE CODE?
Clone the Starter
Prefer to dive straight into working code? Clone the Week 1 branch of the course repo and follow along from there.
AI Engineering · Week 1
11
AI Engineering · Week 1
12
THE PROJECT
What We're Building: An AI Personal Assistant
Here is the service we will build together, starting with Nylas email access and adding intelligence one workflow at a time:
Connects email and calendar accounts through Nylas
Receives real-time email events through signed webhooks
Normalizes provider payloads into typed application models
Uses AI workflows to classify requests and decide next steps
Takes approved email and calendar actions through Nylas
AI Engineering · Week 1
13
THE PROJECT
High-Level Architecture
Email & Calendar
›
Nylas API
›
Signed Webhooks
›
FastAPI
›
Typed Events
›
Assistant Workflows
Nylas sends provider events to FastAPI. The app verifies, models, displays, and routes each event into the assistant workflow.
THE PROJECT
Tech Stack
Python
Pydantic
OpenAI API
SQLAlchemy
Docker
uv
AI Engineering · Week 1
14
PostgreSQL
Pinggy or Serveo
Nylas API
0
PROJECT SETUP
Design First — Draw Your Flow Diagram
Now that you've seen how the AI News API's pieces fit together, do the same for your own idea: before writing any code, sketch your project as an interactive flow diagram — map the pieces and how data moves between them.
Tip: tools like Excalidraw or FigJam work well for this — and Claude can help you talk through the design out loud.
AI Engineering · Week 1
17
1
PROJECT SETUP
Install Python & Git
Install Python 3.10+ and Git — the two foundational tools every project in this course builds on.
$ python3 --version
$ git --version
PYTHON
GIT
AI Engineering · Week 1
18
PROJECT SETUP
RECOMMENDED TOOLING
Set Up Your IDE: Cursor
Cursor is a VS Code–based editor built for AI-assisted coding: intelligent autocomplete, refactoring suggestions, and contextual help are built directly into your workflow. Many engineers find that tools like this meaningfully speed up day-to-day development — it's worth a session to get comfortable with it.
SETUP
AI Engineering · Week 1
19
2
PROJECT SETUP
Install Packages: uv & Docker
uv is a fast Python package and dependency manager. Docker lets us containerize the app so it runs the same way on every machine.
# install uv (see official docs for your OS)
$ curl -LsSf https://astral.sh/uv/install.sh | sh
# install Docker Desktop from docker.com
UV PACKAGE MANAGER
AI Engineering · Week 1
20
3
PROJECT SETUP
OPTIONAL · RECOMMENDED
Install Claude CLI
The Claude CLI lets you work with Claude directly from your terminal — reading, writing, and running code alongside you.
$ claude --version
# follow the official install guide for your OS
AI Engineering · Week 1
22
4
PROJECT SETUP
Give Claude Context: CLAUDE.md
Create a CLAUDE.md file with general context Claude should know about the project — background it needs, but nothing that requires immediate action.
Include: the project's goal, the tech stack, folder conventions, and your preferred coding style.
AI Engineering · Week 1
23
5
PROJECT SETUP
Create the Initial Project Structure
“Create the initial project structure exactly as defined in CLAUDE.md. Create the folders and placeholder files only. Do not implement business logic yet.”
project/
├── src/
├── tests/
├── main.py
└── CLAUDE.md
AI Engineering · Week 1
24
6
PROJECT SETUP
Hello World in main.py
Confirm the basic structure runs before adding anything else.
# main.py
def main():
print("Hello, AI Engineer!")
if __name__ == "__main__":
main()
AI Engineering · Week 1
25
7
PROJECT SETUP
Run uv sync
This creates your virtual environment and a uv.lock file — locking exact dependency versions so the project stays reproducible.
$ uv sync
AI Engineering · Week 1
26
8
PROJECT SETUP
Docker Compose & Docker Run
Start with the simplest possible setup: a Dockerfile that just copies your current code into the image.
$ docker compose up --build
$ docker run your-image
AI Engineering · Week 1
27
9
PROJECT SETUP
GitHub Authentication & .env Setup
GitHub Authentication
Set up SSH or a personal access token so you can push your work. We'll walk through the steps together — or ask Claude/ChatGPT for a guided, step-by-step tutorial.
.env Setup
Copy the example environment file and fill in your own keys (including your OpenAI key).
$ cp .env.example .env
AI Engineering · Week 1
28
10
PROJECT SETUP
Understanding .env.example
If you cloned the repo instead of building from scratch, you'll see a .env.example file — a template listing which secrets and settings the project needs, without any real values filled in. Copy it, rename it to .env, and fill in your own keys there.
# .env.example — checked into the repo (safe to commit)
OPENAI_API_KEY=
DATABASE_URL=
# your local copy — never committed
$ cp .env.example .env
.env is already in .gitignore — only .env.example should ever be committed.
AI Engineering · Week 1
29
11
PROJECT SETUP
Set Up Your OpenAI API Key
You'll need an OpenAI account and API key to follow along with the exercises.
# .env
OPENAI_API_KEY=sk-...your-key-here...
GET YOUR KEY
AI Engineering · Week 1
30
PART 2
Additional Setup
AI Engineering · Week 1
10
11
PROJECT SETUP
Set Up Your Nylas Account: follow instruction here
You need a Nylas developer account so the assistant can connect to a development email account.
# .env
NYLAS_API_KEY=your-api-key
NYLAS_CLIENT_ID=your-client-id
NYLAS DASHBOARD
AI Engineering · Week 1
30
WEEK 1 — RECAP
You're Set Up — What's Next
NEXT SATURDAY
Week 2: The LLM Layer
Connecting your first LLM layer and starting the FastAPI backend.
Bring your laptop, your curiosity, and your CLAUDE.md.
AI Engineering · Week 1
31