Working with Tables

Tower integrates with Apache Iceberg for lakehouse storage. These examples demonstrate reading, writing, and maintaining Iceberg tables.

Writing to an Iceberg Table

This example downloads stock ticker data from Yahoo Finance and writes it to an Iceberg table. The pipeline uses upsert for idempotency—you can safely re-run without creating duplicates.

Highlights: Iceberg, PyArrow, yfinance, upsert

Analyzing Data in an Iceberg Table

This example reads ticker data from Iceberg, computes 7-day and 30-day moving averages plus volatility using Polars, and uses DeepSeek R1 to generate buy/sell/hold recommendations.

Highlights: Iceberg, Polars, Hugging Face Hub, DeepSeek R1

Deleting Iceberg Table Records

This example maintains a rolling window of data by removing records older than a specified time window. The app is idempotent and designed to run on a daily schedule alongside the write app.

Highlights: Iceberg delete operations, scheduling, data retention