Nannyml

NannyML is a post deployment monitoring product for machine learning models. Its core idea is to estimate model performance when ground truth labels are delayed or missing, so a team can see how a model is doing and be alerted when performance drops rather than when data merely drifts. The company says this cuts false alarms, because not every data shift affects results. The cloud product is built on the open source NannyML library. It offers concept drift detection, multivariate and univariate drift detection, prediction and target drift, data quality checks, issue ranking and root cause analysis, plus a cost benefit matrix that ties model performance to business outcomes. It supports batch and streaming data, custom metrics, thresholds, email and Slack notifications, webhook actions and API access. Customers can deploy it in their own cloud, and it is listed in the Azure and AWS marketplaces. It is aimed at data science and machine learning teams, with solution pages for finance among others. The open source library is free. Cloud pricing starts at $399 per month for the Starter plan, which covers 2 models and 10 million predictions with email support, and $999 per month for the Scale plan, which covers 6 models in the customer's own cloud with a private Slack channel. Both include a 30 day free trial. The Enterprise plan has unlimited models and predictions and 24 hour support, and is quoted on request. The site also announces an acquisition of the company.