Home Automation and Pipelines

Automation and Pipelines

The automation need: prompting for correct SQL and pandas, and wiring LLM steps into pipelines that clean, triage, and self-heal.

ADVERTISEMENT

Guides

All guides →

Building a Python Pipeline That Self-Heals Broken Column Mappings With an LLM Step

How to catch renamed and reordered upstream columns automatically without letting a language model silently corrupt your schema.

Dmitri Volkov, a senior analytics automation engineer · 9 min read

Prompting for pandas: Getting Correct groupby Logic Instead of Plausible-Looking Code

Generated pandas aggregations often run clean and return the wrong number; here is how to prompt and verify so the logic is actually right.

Dmitri Volkov, a senior analytics automation engineer · 8 min read

Wiring an LLM Agent Into Airflow to Triage Failed Data-Quality Tests

A read-only triage agent that classifies why a dbt or Great Expectations test failed and routes it, without ever touching your data.

Dmitri Volkov, a senior analytics automation engineer · 9 min read