Exercises

Test your understanding of document processing, ETL pipelines, and DeepSeek-OCR. Progress from warm-up to stretch.

Warm-Up

1. Order the ETL Pipeline

2. What Gets Lost in Flattening?

When traditional OCR flattens a document to text, which are lost? Select all that apply.

Core

3. Complete the Architecture

4. Compression vs Accuracy

Find the configuration that achieves >95% accuracy for financial documents.

5. Token Cost Visualizer

Drag the slider to change batch size. Watch how token counts and costs compare across VLM systems in real time.

Stretch

6. Match Tool to Layer

Which tool provides the most leverage at each pipeline layer?

7. Visual Document Router

Configure a document's properties, then click Route to see which OCR system it gets sent to and why. The path animates through the decision tree.