AI-rflow Meetup: Exploring AI Use Cases in Apache Airflow
Details
Join fellow Airflow enthusiasts and leaders for a special Airflow Meetup focused on AI Use Cases in Apache Airflow.
We'll start you off with a deep dive into building an AI-Native Data Platform, and finish off with a session on the AI-Airflow feedback loop.
Come for the presentations, stay for the food, drinks, and swag!
PRESENTATIONS
Talk #1: Building the AI-Native Data Platform: How agents reshape build, serve, and compound insights from data
- Speaker: Jordan Kail, Data Engineer at Together AI
AI is changing how data platforms are designed and operated. As intelligent agents become more capable, they’re beginning to play a larger role in how teams build pipelines, serve insights, and continuously improve the value extracted from their data.
In this talk, Jordan from Together AI will explore what it means to build an AI-native data platform. One where agents help automate development, accelerate experimentation, and compound insights across systems. We’ll discuss how organizations can rethink traditional data architectures to support agent-driven workflows, scalable model deployment, and faster iteration on data products.
Attendees will walk away with a clearer picture of how AI agents are reshaping modern data infrastructure and what it takes to prepare your platform for this shift.
Talk #2: The Airflow–AI Feedback Loop and the Technology It's Inspired
- Speaker: Steven Hillion, SVP of Data & AI, Astronomer
Over the past two years, we’ve seen two concrete shifts in how leading data teams use Apache Airflow.
First, Airflow is increasingly orchestrating AI workloads: large-scale batch inference, content embedding, and evaluation pipelines including human-in-the-loop reviews. In many organizations, Airflow now coordinates not just ETL, but model lifecycle operations and agentic workflows.
Second, as AI systems become responsible for generating and modifying data pipelines, Airflow’s metadata layer — DAG structure, task state, lineage, ownership, and historical runs — becomes high-value context for AI agents. Airflow is no longer just an orchestrator; it is a structured control plane that AI systems can reason over.
This creates a feedback loop:
- Airflow orchestrates AI workloads.
- AI agents use Airflow metadata to generate, validate, and evolve pipelines.
- Human-in-the-loop patterns close the control gap.
We’ll walk through examples of this pattern and discuss the tooling it has inspired, including Astronomer's AI agent tooling, the Airflow AI SDK, and human-in-the-loop operators.
AGENDA
- 5:30-6 PM: Arrivals, networking, food & drinks
- 6-7PM: Presentations
- 7-8PM: Networking
