September 12, 2023 | Must Reads

A Radically Different Vision for Transforming and Analyzing Data: Introducing Feldera

Written by

Tony Liu

Once I understood the Feldera founders’ vision, everything clicked. The technology that the team had built had the opportunity to upend the way we think about ETL, ELT, analytics, and the Modern Data Stack. 

The fundamental insight was this – instead of moving data from one place to another and then transforming it there, what if we transformed it before sending it to the destination? That’s when I realized that this was not just about building an analytics tool for data in motion but really about changing the way we do transformation and analytics.

The current model involves pulling data from some source via Fivetran or Airbyte into Snowflake or Databricks and then running various dbt models to transform the data, each model resulting in tables being stored and loaded. It is easy to take this model for granted since it is ubiquitous and has become the standard. 

On closer inspection, it becomes clear that there are many opportunities for improvement at each step of the workflow. As a firm, we met many companies that were looking to improve one or two of these steps. When I met the Feldera team, it soon became clear to me that their radically different vision was the one that could result in a step function improvement in the way companies transform their data both at rest and in motion. 

Leonid, Mihai, Ben, Lalith, Gerd, and Chase were all members of VMware Research. Mohsin, who is the CEO, is the only team member who was not part of the research lab (and also the only one without a PhD). Instead, Mohsin was the technical lead for VMware Skyline and the first customer for the technology that the rest of the team had been developing. Together they were able to implement the previous version of Feldera’s underlying technology, Differential Datalog, into Skyline and achieve improvements in performance that seemed unbelievable to customers. The learnings from that work inspired their DBSP algebra, which is the basis of Feldera’s open-source engine and recently won Best Research Paper at VLDB 2023.

Aside from their deep technical chops, what makes the Feldera team special is that they are starting the company untethered to standard technologies and workflows in data analytics. Their outsider status has actually been a strength. They are able to work from first principles and observe many of the practices many of us had taken for granted with a fresh and inquiring set of eyes. 

Perhaps most importantly, they are acutely attuned to the needs and desires of both developers and engineering leadership. They know that great technology is just that. Great technology. Which is why they are determined to build a product that makes a difference for businesses, one that can accelerate the bottom line growth of companies. They see it as similar to how adoption of DDlog fundamentally changed what was possible for VMware Skyline. But even better, now any company can reap the benefits of what they’re building.

We at Costanoa are thrilled to be leading Feldera’s $6 million seed round. I’ve been really impressed by their progress so far and look forward to continuing the close partnership we’ve enjoyed with them since day one of their becoming a company.


Our Perspectives