Big data transformation startup Tobiko Data Inc. revealed today it has closed on two separate funding rounds, including a $17.3 million Series A round that followed an initial $4.5 million seed raise.

The Series A was led by Theory Ventures and saw participation from 20Sales, plus the chief executive officers of Fivetran Inc., Sutro Labs Inc and MotherDuck Corp., while Unusual Ventures led the seed round. The total amount raised via the two rounds was $21.8 million, and the announcement coincides with the launch of its managed cloud and enterprise platforms, which are both generally available starting today.

Tobiko Data has created a data transformation platform that’s built on top of the open-source SQLMesh and SQLGlot projects. It’s designed to help companies easily build more efficient and scalable data pipelines for artificial intelligence and other workloads.

SQLMesh is a tool that applies DevOps best practices to data pipelines written in SQL or Python. It helps data teams run, test and deploy data transformations efficiently and safely with features like virtual environments, automatic DAG generation and smart change categorization. Data scientists and analysts can create data pipelines using the familiar Structured Query Language, with minimal involvement of data engineers, reducing much of the manual toil involved with such tasks.

Meanwhile, SQLGlot is a key component of SQLMesh that acts as an SQL parser, transpiler and translator, with support for 24 different SQL dialects.

According to Tobiko Data, many enterprise data teams believe they are shackled by the lack of scalability offered by existing data transformation tools, which still require too much manual work and suffer from inefficient compute.

Co-founder and CEO Tyson Mao said existing data transformation products are fundamentally broken because they’re unable to make sense of SQL. Because of this, he said, they blindly execute scripts without any comprehension of the underlying changes that result.

Mao said this is a problem because a single change in a query can “affect billions of rows of data.” He added that some companies “spend millions every year on unnecessary rebuilds of the warehouse when only a small precise change is needed.”

The startup offers an alternative that doesn’t waste costly and unnecessary compute cycles on the data warehouse, which can take hours or even days, depending on the size of the company and how much data it owns. Tobiko simplifies data pipeline creation because its platform has a better understanding of SQL semantics, which means it will only execute the necessary downstream changes, rather than completely rebuilding the data warehouse each time a new data pipeline is added.

It also supports virtual data environments that can understand and remember every version of every data pipeline, so it can avoid duplicative computation. “Tobiko eliminates redundant computations, which can become extremely costly,” Mao said.

The startup said the data integration company Fivetran has already transformed its data transformation approach after implementing its platform. Fivetran CEO George Fraser said his team was beyond impressed with the capabilities of SQLMesh.

“We’ve been using various tools to power our analytics and started building our own in-house features,” said Fraser. “When we came across SQLMesh, we realized it has the advanced features we need, but on steroids, and more. We love the product’s innovative approach to virtual data environments and column-level lineage-based categorization.”

Tobiko Data said it will use the funds from today’s round to continue developing SQLMesh and SQLGlot. The company plans to showcase its platform at upcoming events, including the Snowflake Summit on June 6 and Databricks Data + AI Summit on June 10.

Image: SiliconANGLE/Microsoft Designer