Arize AI is a company developing a machine learning (ML) observability platform for real-time monitoring, analysis, and explainability. ML observability is the practice of understanding a model's data and performance across its lifecycle. It enables ML developers to identify the root causes behind a model's behavior in order to improve it. The Arize platform helps developers monitor, troubleshoot and explain models. It is used by data science and ML engineering teams to perform the following:
- Monitor real-time model performance, with support for delayed ground truth/feedback
- Perform root-cause model failures/performance degradation using tracing and explainability
- Conduct multi-model performance comparisons
- Surface drift, data quality, and model fairness/bias metrics
- Log model inferences across training, validation, and production environments
Diagram showing Arize functionality and how the platform fits into ML workflows.
Arize acts as an inference/evaluation store to better understand ML models once they are deployed into production. It is an open platform that works within the user's ML infrastructure and can be deployed as SaaS or on-prem. Atrize natively supports:
- Tabular/structured data types (strings, floats, booleans, etc), as well as embedding support for NLP, image, and other unstructured data types.
- Binary classification, multi-class classification, regression, ranking, NLP, and CV model types.
The model type informs the data ingestion format and the performance metrics that can be utilized in the platform. Arize surfaces outlier/anomalous data through data quality checks, drift checks, and feature performance heatmaps.
Arize AI was founded in 2020 by Jason Lopatecki (CEO), Aparna Dhinakaran (CPO), Michael Schiff (CTO), Mikel King (Technical Lead), Gabe Barcelos (Head of Solutions Architecture), Tsion Behailu (Engineer), Manisha Sharma (Engineer), and Nate Mar (Engineer). The company was formed through the common frustrations of Lopatecki and Dhinakaran when it came to understanding ML model behavior. This led to the development of the Arize platform for ML observability.
Arize AI announced it had already raised $4 million in seed funding when it came out of stealth on February 18, 2020. The funding was led by Foundation Capital, with participation from Wilson and Trinity Ventures. On September 28, 2021, the company announced $19 million in Series A funding led by Battery Ventures. Other participants included existing investors Foundation Capital, Trinity Ventures, The House Fund, and Swift Ventures. On September 8, 2022, Arize AI announced a $38 million Series B round led by TCV with participation from Battery Ventures and Foundation Capital.
