Part 2 — Python & AI

Part 1 got you a working setup. This part is what you do with it: a Python environment that travels, a key kept out of git, and a script that sends a real SEC filing to a language model and validates what comes back.

10:30 to noon. A short lecture on Python environments and Stanford’s AI services, then the second work block. Finish with the extraction checkpoint: turn one public filing into a validated record and save your work to your fork.

This part assumes the Part 1 Checkpoint passed. If any of its four items is still broken, fix that first — a venv you cannot create or a fork you cannot push to will stop you here rather than politely waiting.

---

Hands-On Lab

Work through them in order — each builds on the one before.

Section	Format	What you’ll learn
Running Python on the Yens	💻 Hands-on	How $PATH decides which python3 answers, and the three ways to run Python
Python Environments	💻 Hands-on	Build an isolated venv and rebuild a whole project from its requirements.txt
AI Services & Data Privacy	💬 Demo + discussion	Try the Playground, explore Stanford’s AI services, and check what data each service can receive
Managing API Keys	💻 Hands-on	Load ANTHROPIC_API_KEY from .env, verify it safely, and prove Git ignores it
Extracting Data with an LLM	🔑 Checkpoint	Extract and validate a filing, research the person with web tools, and save a summary with its source

The extraction checkpoint brings together the skills from Part 2. Day 2 uses the same script for profiling and Slurm jobs. If a checkpoint step fails, ask an instructor for help before moving on.