Solving the Data Engineering Bottleneck in Production ML

If your ML projects are moving too slowly, it's likely because complex data engineering is standing in the way. ML teams get bogged down by building the data pipelines and serving infrastructure needed to develop and deploy new features for production.

In this 20-minute talk, David Wang, Technical Evangelist at Tecton presents a new approach to the overall data and feature workflow for AI/ML. So that, instead of spending engineering time setting up pipelines, ensuring data consistency, handling feature sprawl and countless other data infra-related tasks, ML teams can simply write features as code via a declarative framework.

David delves into real-world scenarios, such as building a real-time recommender system, best practices and key considerations for streamlining the data engineering for ML, and introduces a simpler way to approach the data and feature workflow (which has driven an 80% improvement in model time-to-production).

What you’ll learn:

• How Plaid, HelloFresh, Coinbase accelerate model development and production
• Key best practices and considerations for streamlining feature production and serving
• Overview of Tecton and how its declarative framework drives faster dev and better models