(Day 350) Using Airflow for a RAG app

Ivan Ivanov · December 16, 2024

Hello :) Today is Day 350!

A quick summary of today:

  • covering week 4 - Streaming from Zach Wilson’s YT camp
  • streaming an amazing free course on ‘GenAI with Apache Airflow’

Streaming by Zach Wilson

Streaming with Kafka and Flink P1 Streaming with Kafka and Flink P2 Streaming with Kafka and Flink P3 Streaming with Kafka and Flink P4 Streaming with Kafka and Flink P5

Stream - GenAI with Apache Airflow course

I covered [this(https://academy.astronomer.io/introduction-to-genai-with-apache-airflow/1986175] amazing free course on Astronomer on using Airflow for a RAG app. It was amazing, and learned a lot about how Airflow works and making a more complex pipeline like:

image

The 2nd DAG was just to delete the docs from the db:

image

The UI looks like:

image

and:

image

I have thought about an app like that for this blog as well. However, questions like: “what did Ivan study” would have to force a traverse of the whole blog’s info, which can just be done with a preset answer 😆 which I will kind of write for the last day. And questions like “what did Ivan study in day X” - well that might be a slight use case, but also can easily be answered even without an LLM in the RAG - just return the summary I write at the start of each post ~ Nevertheless it was a great blog and taught me some new things about Airflow like dynamic (parallel) task execution, branching, and just creating a more complex DAG.


That is all for today!

See you tomorrow :)