Thierry Cruanes
Co-founder and CTO of Snowflake.
Session — September 30, 2021
From 1st Schema Then Data, To 1st Data Then Schema
A discussion about inventing new ways of organizing data, unlocking patterns in the chaos, and where “1st data then schema” will take the world of data next.
About the session
Humans as pattern recognition machines
Humans have evolved to be pattern recognition machines. This is why we see images in the clouds, but it also helps us function in daily life — and it explains why people feel uncomfortable and overwhelmed by chaos.
Data is no longer a tidy set of tables waiting to be catalogued. Modern pipelines move streams of events, records, and signals whose shape is often decided long before anyone writes a schema. This session explores the idea of arriving at the schema after the data itself, rather than before it, and what changes when organizations treat structure as something discovered rather than declared.
The conversation focuses on how teams invent new ways of organizing information, unlocking patterns in what first appears to be chaos, and building architectures that can evolve as quickly as their sources. Attendees considering a schema-first or data-first approach hear how practitioners weigh flexibility against governance, and how early decisions ripple through pipelines, platforms, and downstream consumers.
Humans have evolved as pattern recognition machines, which is why we see familiar shapes in clouds and why unfamiliar data feels unsettling. That same instinct, applied deliberately, can reveal structure in noisy feeds. The talk connects this human capacity with the engineering techniques that let teams surface patterns at scale, then turn those insights into repeatable, well-designed data flows.
Whether you are a developer, data engineer, or architect working with streaming pipelines and cloud platforms, this session offers a perspective on where data-first thinking is headed. It encourages attendees to question assumptions about structure, share approaches that have worked in practice, and consider how flexible organization can reduce friction for the professionals and teams who rely on data every day.
When there is too much “noise” (data of any kind), we must develop technologies to modify it to become meaningful. In this way, humans and information co-evolve, and the machines we build accelerate subsequent generations’ ability to learn faster with more fidelity.
This session brought together two innovative builders — Thierry Cruanes, co-founder and CTO of Snowflake, and Chris Messina, inventor of the hashtag — in a discussion about inventing new ways of organizing data, unlocking patterns in the chaos, and where “1st data then schema” will take the world of data next.
| Title | From 1st Schema Then Data, To 1st Data Then Schema |
|---|---|
| Date | September 30, 2021 |
| Time | 08:00 AM – 08:30 AM |
| Speakers | Thierry Cruanes, Chris Messina |
Speakers
Two builders on data and patterns
Chris Messina
Inventor of the hashtag.
Where to go next
More from DataOps Summit
Agenda
Conference Agenda
Keynotes, breakout sessions, and training days across the summit program.
Events
Digital Transformation in Healthcare
How Humana uses DataOps to redefine the senior experience through integrated care delivery.
Join us at the summit
DataOps Summit is hosted by StreamSets and brings together data engineers, data architects, data scientists, and technology leaders for keynotes, breakout sessions, and hands-on training.