End-to-end solution for deploying and maintaining your machine learning models in a production environment


Model as a Service (MaaS) is a concept that has been gaining traction in the data science community in recent years. The idea behind MaaS is to provide you, the data scientists, with an end-to-end solution for deploying and maintaining your machine learning models in a production environment. MaaS abstracts away many of the technical details involved in deploying machine learning models. This can free up your time to focus on what you do best: building and running models that deliver meaningful insights and predictions.

For data scientists, the key to successfully using MaaS is to be aware of its strengths and limitations and to use it in conjunction with other tools and techniques. For example, while MaaS can simplify the deployment process, data scientists still need to be involved in testing and validating their models.

Additionally, MaaS solutions are not a replacement for traditional data science tools and workflows. Instead, they are complementary tools that can help to streamline the deployment process and make it more efficient.

The success of a MaaS solution depends on several factors, including the quality of the models being deployed, the availability of high-quality data, and the infrastructure that supports the deployment process. MaaS helps streamline the deployment process of your machine-learning models. Combining your models and our execution platform is the recipe for predictive success.

The Challenge

Develop and Train

  • Data science teams are very good at picking up data, train models and apply them to test sets
  • Great to showcase the value & business case
  • Not ready for full deployment with continuous operations

Execute and Refine

  • For operations, an execution framework to deploy, operate, collect model feedback and retrain the model is required
  • The Framework is also ensuring data and model governance & life cycle management, consistent model inference, and auditability

Figure 1. Squirro’s MaaS focuses on the technical details so that you can focus on delivering valuable insights for your organization.


Squirro's Model as a Service, platform is a comprehensive solution for data scientists who want to deploy their models in a production environment. Here's how it works:

Ingesting Data

Squirro ingests data from various sources, including databases, APIs, and file systems. Data scientists can select the data sources they want to use for their models and configure the ingestion process to ensure that the data is properly structured and formatted.

Appending Security

Squirro takes security very seriously, so all data is securely stored and processed. Data scientists can set up access controls to ensure that only authorized users can access the data and models.

Fetching Models

Once the data has been ingested, data scientists can fetch their models and integrate them with the data. Squirro provides a simple interface for uploading models, so data scientists can easily get started with the deployment process.

Inference Job

Squirro runs an inference job on the data and models, generating predictions and insights. The results of the inference job are stored in a centralized repository, so data scientists can easily access the output and use it to build end-user applications. Both CPU and GPU-based inference jobs are available.

Building End-User Applications

Data scientists can use the output of the inference job to build a wide range of end-user applications, including dashboards, reports, and alert systems. Squirro provides various tools and templates to make it easy for data scientists to build applications that meet the needs of their stakeholders.

To summarize, Squirro's Model as a Service platform provides data scientists with a comprehensive solution for deploying their models in a production environment.

Figure 2. Combining your models, or models from libraries, with our platform to ensure smooth deployment and running of the models.


Squirro is a cutting-edge Model as a Service (MaaS) platform that offers a wide range of benefits for data scientists. Here are just a few of the reasons why Squirro is particularly well-suited for model deployment:

Model as a Service

Develop your models in your own environment(s) inside your company or use models available on platforms such as Hugging Face.

Squirro allows for inference jobs based on CPU or GPU-based infrastructures. You can also direct Squirro to use your inference infrastructure and write the output back to the Squirro platform to be used in applications.


Squirro integrates with a wide range of enterprise data sources, so data scientists can deploy models that take advantage of a variety of data sources and data types. The results of your model jobs may be used in your 3rd party applications, injected into Apps you build yourself on the platform, or used in some of the ready-to-use apps provided for marketing, sales, service, and risk.


Squirro is built to scale, so data scientists can deploy models that can handle large amounts of data and complex use cases. Whether you're deploying a model for a small team or a large organization, Squirro has you covered.


Squirro's fast and efficient deployment process means that data scientists can get their models up and running quickly, so they can start getting insights and predictions right away.


Squirro's robust security features ensure that your models and your data are secure and protected. At ingestion time, full ACL compliance is ensured. Data scientists can be confident that their models are deployed in a secure environment.

User-friendly interface

Squirro's user-friendly interface makes it easy for data scientists to deploy their models and manage their deployments. With a simple drag-and-drop interface, data scientists can easily upload their models and integrate them with other data sources and tools.

Squirro is an outstanding Model-as-a-Service platform that offers data scientists a powerful and easy-to-use solution for deploying their models.



The Challenge

Squirro’s MaaS solution has already proven effective in multiple cases, one of which is regarding case classification, where several million cases of customer requests had to be classified. This often resulted in slow responses, unhappy clients, and frustrated support teams. Ultimately, costly remedies would have to be used to compensate for the lack of automation:


The Solution

Squirro’s MaaS solution offered a fast and reliable way to automate case classification with more than 90% accuracy. The resolution handling was also drastically improved by, amongst others, automatically providing the right resolution documents to the right person based on previously resolved similar cases:


Additional Benefits

Squirro’s MaaS solution came with additional benefits beyond just the execution of the case support classification and regular handling. It was able to provide insights that would have otherwise been left unutilized. Amongst these was the automated alerting of line managers and relationship managers if the request were business critical and required extra attention. Another insight came from the clustering of repetitive complaints. This exposed operational issues that would have otherwise been harder to pinpoint and immediately provided well-documented proof:

Next Steps

If this brief product overview has made you curious to what extent Squirro’s MaaS solutions can benefit your organization, please fill out this form. One of Squirro’s MaaS product experts will contact you to discuss the possibilities tailored to your needs.

This site is registered on as a development site.