Tecton: Interpreting machine learning data into operational features

An Intellyx Brain Candy Brief

Tecton works within the seismic rift between massive event flows and data lakes, and the AI/ML platforms that are being driven to provide meaningful insights and responses from that data.

Provisioning features, which are actionable instruction sets of data for machine learning platforms such as Google Cloud AI, AWS SageMaker, and other interpretive analytics tools is no small feat, especially when you consider the scale and variety of real-time and archived data sources at hand. The founders of Tecton emerged from the Michelangelo ML training project, which powered many of the optimization features under the hood at Uber.

Teams can easily spend most of their feature engineering time writing Python code on their local laptops, then building production pipelines for online serving, rather than focusing on models for predicting traffic patterns, or scheduling deliveries, or selecting the best available person to respond to a support issue.

Tecton’s solution allows ML teams to use a Python SDK to create feature transformations combining data from different sources such as data warehouses, data lakes, event-based streams including Kafka and Kinesis, and real-time data passed at prediction time. The results are published into a self-service ‘feature store’ of definitions the AI/ML or analytics platform of choice can pull from via API.

©2020 Intellyx, LLC. At the time of writing, Tecton is not an Intellyx customer. Want to see more BrainCandy? Subscribe today. If you are a vendor seeking coverage from Intellyx, please contact us at PR@intellyx.com.