The Story of Mage: Building the Future of Data Infrastructure

Sometimes the biggest opportunities come not from building something entirely new, but from making existing processes dramatically better. In a recent episode of Category Visionaries, Tommy Dang shared how his experience at Airbnb led to the creation of Mage, a company reimagining how developers interact with data infrastructure.

The Airbnb Years

In 2015, Tommy joined Airbnb when it was still relatively small. “There’s less than 200 engineers and the company was less than 2000 by today’s standard,” he recalls. The company’s core value of being a “cereal entrepreneur” shaped his approach to problem-solving: “You treat the company as an owner, so you take it upon yourself to do what you got to do to hit goals, to help users to help customers.”

This environment of ownership and flexibility allowed Tommy to explore different areas of the business. “You got to do just literally anything you want. You can move to any team, work on any project as long as it helps achieve our company’s mission,” he explains. This freedom led him to focus on data-intensive tooling and products, addressing significant gaps in the company’s infrastructure.

The Spark of Innovation

Working with data tools at Airbnb, Tommy identified a crucial gap in the market. “We had disparate tools for pulling data, extracting data, transforming data, storing the data, building training sets, reusable data sets, feature engineering, also deploying models, training machine learning models,” he remembers. While these tools served specialized roles like data scientists and engineers, there was an opportunity to make them more accessible to a broader technical audience.

The Pivot

In December 2020, Tommy left Airbnb to start Mage. The initial vision was to build an end-to-end machine learning platform. They launched in early 2022 with paying customers and thousands of sign-ups. However, they soon discovered a more fundamental need in the market.

“What we found is they actually struggled with a more urgent data challenge early in their journey. And it is just the movement of data, the transformation of data, the integration of data,” Tommy explains. This realization led to a pivotal decision: they took their core data pipeline technology and open-sourced it.

Breaking Through

The open-source strategy proved transformative. “We open sourced that earlier this year, and it’s been catching on like wildfire,” Tommy shares. Within months, they had over 2,000 GitHub stars and a thriving community of 400-500 Slack members.

Their success came from focusing on four key differentiators from existing solutions like Airflow. First was developer experience – making the tool enjoyable to use. Second was building in engineering best practices, ensuring modular and testable pipelines. Third was treating data as a first-class citizen with built-in versioning and partitioning. Finally, they focused on making scaling simple, even for small teams.

The Future Vision

Looking three years ahead, Tommy sees Mage becoming an essential part of every company’s data infrastructure. “We see a world where Mage is the go-to data tool for early stage companies, mid sized companies. It’s a tool that you think about and you spin up as soon as you start a company, as soon as you have any database set up.”

But perhaps most telling is how Tommy envisions Mage’s role: “We love doing being the dirty and boring plumbing behind the scenes for companies… We want to get to a place where everything is so easy, so smooth and so transparent that you even forget that we’re here.”

This vision – of becoming invisible infrastructure that just works – reflects a deep understanding of what developers truly want: tools that solve their problems without creating new ones. In a world where data infrastructure is becoming increasingly critical, Mage is betting that the future belongs to tools that prioritize developer experience and reliability above all else.