Data observability startup Telmai Inc. today said it has raised $5.5 million in an oversubscribed seed round, bringing its total funding to $8.5 million.

The company has created what it says is a unique approach to data quality management with an engine based on the Apache Spark analytics framework. It can accommodate a wide variety of data types and data stores while calculating metrics within the data pipeline rather than after it has been committed to a destination.

The platform works with structured, semistructured and event-streaming data sources in data lakes, cloud data warehouses and on-premises relational stores. Telmai said its platform is highly scalable and can detect data issues across large volumes of diverse data at marginal cost.

“Our architecture is designed to work with any system, whether it’s batch, a data lake or Delta Lake and with data formats in Parquet, nested JSON, semistructured data or streaming and calculate the metrics in a decoupled manner,” said co-founder and Chief Executive Mona Rakibe (pictured). “We are able to support all of these different systems without any prepping of data at scale in a very easy-to-use manner.”

Data diversity

The system’s platform independence is in line with the increasing diversity of analytics platforms businesses are using, she said. “Even small companies have a myriad of systems underlying certain use cases,” she said, where they might be using Amazon Web Services Inc.’s Redshift or Athena for certain use cases along with Google LLC’s BigQuery, Microsoft Corp.’s Azure Blob and Databricks Inc.’s Delta Lake. “When they have such heterogeneous systems, they are building their own pipelines based on what solves their use case best,” she said. Telmai aims to obviate roll-your-own solutions.

Another unique attribute of the Telmai platform is that it can detect anomalies at the value level, according to Rakibe. “A lot of tools are designed for data warehouse; they look at metadata and metrics around the schema and things like that,” she said. “We look at the values themselves and figure out if there are anomalies or outliers, such as sudden pricing changes or changes in revenue metrics.”

The platform classifies data and presents it to users “in an easy-to-understand user interface that ties back into data quality key performance indicators,” she said. “We have even heard from customers that they thought of Telmai as an educational experience for teams that need an understanding of how they should be thinking about data observability and how that translates into data quality.”

Machine learning is applied to adjust thresholds automatically as data volumes change. Because the engine isn’t tied to a data store, it can scale and operate at high speed, according to Rakibe. “We are not running queries on [a data warehouse],” she said. “We actually read the data and calculate the metrics in our Spark-based engine.” Identifying data quality issues close to the source of the data also reduces costs.

Strong start

Although the company is just two years old, it has signed up Datastax Inc., Merkle Inc. and API Hub Inc., which does business as Clearbit, as reference customers. Telmai was looking to raise less than the $5.5 million it closed but “our existing investors were interested when they started seeing the signals in the market,” Rakibe said. The company has just 14 employees but has a joint marketing agreement with Google Cloud and lesser partnerships with Snowflake Inc. and Databricks. 

“We were closing customers at a good [annual recurring revenue rate], even though we were extremely small,” Rakibe said.  “We were capital-efficient and we had built a product with a clear differentiation and a clearly defined value proposition. I feel fortunate.”

Telmai will use the funding for demand generation and to grow its employee base. The funding round was co-led by Glasswing Ventures LLC and .406 Ventures LLC, with participation from current investors, including Zetta Venture Capital LLC Partners.

Photo: Telmai