About the Chronicles

DataAI Chronicles started as a way to write down what actually happened on the platforms I work on: why a design was chosen, what broke, what it cost and what we changed. Most writing about data and AI describes tools. I wanted to describe decisions.

I am Ved Prakash, a Principal Data Engineer at GitLab. I lead Project Siphon, GitLab's real-time change data capture platform built on PostgreSQL, NATS JetStream, Apache Iceberg and Snowflake, and I designed the company's governance framework for Snowflake Cortex AI. I am a multi-year Snowflake Data Superhero and one of 17 members of the global Data Superhero Council.

The Chronicles has three parts. The writing, here and on Medium. The talks, at conferences such as Snowflake Summit and Berlin Buzzwords. And consulting, for teams that want the same thinking applied to their own platform.

The thesis

Durable models on variable compute. Engines, vendors and AI models change every year. The way you model your data, define your business and govern access should not have to. Most of what I write comes back to separating the parts that must last from the parts that are meant to change.

My full background, case studies and CV are on vedprakashai.com.