Data transformation startup Coalesce emerged from stealth mode on Jan. 20 with $5.92 million in seed funding and its flagship platform, which is now generally available.
Based in San Francisco, Coalesce has been building its data transformation technology since 2020 in an effort to create a more intuitive and automated approach for getting data into the right structure and format for data analytics.
The initial release of the Coalesce data transformation platform is focused on helping users get data into the Snowflake cloud data platform. Coalesce is entering a competitive market with multiple vendors and technologies, including Matillion and dbt (data build tool).
In this Q&A, Armon Petrossian, CEO and co-founder of Coalesce, discusses the challenges and opportunities of data transformation.
Why is Coalesce emerging from stealth now with its data transformation platform?
Armon Petrossian: What we have found is that the bottleneck for every organization trying to be data-driven today is data transformation. What do you do once the data is landed in your platform and how do you prepare it in a way that's consumable from its raw format? In our view, there really isn't a good approach, which is what drove us to build Coalesce and be in stealth mode for as long as we have been, with a focus on building the product and what we want to accomplish.
Armon PetrossianCEO and co-founder, Coalesce
The reason why we came out of stealth now is because the product is ready. There are many complexities associated with transforming data, and there are many nuances in order to offer something that is ready for the enterprise -- and it just takes time.
What do you see as the differentiator for the Coalesce data transformation approach?
Petrossian: We have what we call a column aware architecture.
So the concept of column aware metadata is core to the foundation of our product and our architecture. That is what unlocks everything for our platform, including SQL automation and column-level data lineage, which is something that every organization is struggling with today.
People have traditionally been applying data transformations to data tables. The reality is that it is columns and rows that people deal with every day.
With Coalesce, we are able to identify column metadata from the moment it lands in the platform and use it to automate SQL, automate the construction of dimension tables or automate lineage. Column metadata for us is core to data engineering and data transformation.
What do you see as the most common types of data transformations?
Petrossian: It really depends on the user and their data warehouse architecture. It could be transformation to star schema or Data Vault 2.0. It could be really anything that the user is trying to accomplish with a data project.
Data transformation is really taking data from its unstructured, semistructured or raw formats and transforming it to the point where it's consumable at scale.
Editor's note: This interview has been edited for clarity and conciseness.