Python & Data Toolkit Refresher

Practise and Check

Deliverable

Data warm-up: load a CSV into pandas and into SQLite, answer five questions both ways, and expose one of them through a small Flask endpoint.

Assessment

Answer the five Interview questions without notes, then run the lab.

Interview probe

Larkspur's nightly evaluation sends 200 golden questions to the model. A teammate rewrote it with `asyncio.gather` to make it faster, and it takes exactly as long as before. What do you look for?

Common mistakes

• Normalising rows with np.linalg.norm without keepdims=True
• Building SQL with an f-string instead of a ? placeholder
• Calling blocking llm.invoke inside async def and expecting concurrency

Related Modules

• T2.2LLM Provider APIs & Model Selection
• T2.3Hugging Face Models & Gradio Demos