Snowflake is a cloud computing-based data platform company founded in 2012. Since then, Snowflake has seen huge success and now boasts a large and growing customer base and legion of technical fans.
Like all data solutions, there is a question that remains after placing your data into the platform: How do I unlock the value in my Snowflake data?
With so many customers using Snowflake, we wonder how many have received a quick answer to that question. You can clean, enrich, and validate your data before ingestion or even during ingestion using a variety of methods, but that may not always be a viable option.
Another question commonly asked is: Can I perform actions on my data once it arrives in Snowflake? Absolutely!
One of the best ways to do so is to use artificial intelligence and machine learning (Ai/ML) to validate, enrich or update your data. ElectrifAi’s machine learning models support this through two options: User Defined Functions (UDF) and API Service Calls.
User Defined Functions (UDF)
User Defined Functions (UDF) can be used to deploy ElectrifAi machine learning models for a wide variety of business functions in Snowflake. The UDF option allows you to call machine learning models from within the Snowflake environment based on activity triggers such as inserting or altering a record in a Snowflake database.
Until recently, Snowflake UDF’s only supported Java. This limited the options for companies that standardize on Python variants for machine learning development. The recent Snowflake announcement of support for Python, however, will greatly expand the options for using machine learning models in UDF’s.
The following diagram demonstrates how data is loaded into Snowflake’s data warehouse and then immediately enriched through an ElectrifAi Machine Learning Model deployed as a Snowflake UDF. The UDF output is generated for easy viewing on an AWS Quicksight Dashboard.
API Service Call
With ElectrifAi’s Machine-Learning-as-a-Service (MLaaS) you can call a trained and deployed model from within Snowflake using their API service. The functionality is like calling a UDF on a CRUD event, but the main difference is that the machine learning model is managed outside of Snowflake instead of inside Snowflake.
The following diagram explains how ElectrifAi’s trained machine learning model is accessed from Snowflake through an Amazon Sagemaker managed service using an API service call.
The choice of whether to use the UDF or API Service Call options can vary depending on your own data stack. You can also use a mixture of both based on a variety of factors including the client’s technology maturity, understanding of UDF’s and APIs, and considerations around specific use cases.
ElectrifAi’s machine learning solutions can be utilized for faster time to value and leveraged for Snowflake data lakes using either method.
Using UDF’s or APIs, you can move the power of machine learning close to the most valuable Snowflake data lakes. You can also validate, enrich, and make other machine learning models truly event-driven based on Snowflake data changes.
This will revolutionize how many customers utilize their data with machine learning to significantly increase business value.
If you would like to learn more about how ElectrifAi can help you unlock the value in your Snowflake data, reach out to us today!