New York-Based Machine Learning Startup
If you’re an entrepreneur in New York, there’s a good chance you’ve heard about Wallaroo, a new platform that’s been developed in partnership with Microsoft. It’s designed to help businesses of all sizes, in any industry, get started with building their own mobile applications. The company’s Founder and CEO are a former Microsoft employee, and his model is simple: he’s taking advantage of Microsoft’s powerful cloud computing platform and leveraging it to build a tool that helps his clients find and hire great people.
Founder
Table of Contents
Wallaroo is a cloud-based AI model management and data-processing platform. It was designed as a one-stop shop for data scientists looking to build, deploy, and manage machine learning models. As a result, it is used by leading brands around the world for a variety of purposes, including marketing, fraud detection, customer experience, and more.
The company’s flagship product is the Wallaroo engine, a highly performant data processing engine that is able to run on premises or in the cloud. It offers a number of features, from pre-processing to in-memory computing. Among its many notable features is an observability suite that delivers both compute and model performance metrics. These include an audit trail that enables data scientists to see how their ML models perform in the real world.
Wallaroo’s observable solutions include a data connector, a scalable data processing engine, and a set of MLops (machine learning operations) to enable collaboration between data scientists. Moreover, it boasts a set of features that allow users to create custom integrations with in-house solutions and enterprise data sources. Ultimately, Wallaroo’s patented data management solution combines MLops with a distributed processing engine to create a highly performant platform for running production ML models. In fact, it beats the competition in the ML Olympics when it comes to analyzing tens of thousands of events per second on a single server.
Wallaroo recently announced a $25 million round of funding that will be used to expand its offerings, add more people to its team, and develop its own software.
Platform
Wallaroo is an AI model management and deployment platform that allow data scientists to run ML models on live, real-time data. It also offers tools for observing and comparing the performance of a machine learning model. Using a single API, the Wallaroo platform combines machine learning operations, develops, and distributed processing to deliver the fastest and most flexible production ML environment. This includes a Python SDK interface and a web dashboard.
The Wallaroo data processing engine can run in the cloud or on-premises, and it has a variety of features that include pre-processing tasks, in-memory analytics, and compiling to native code. Its data connectors support a variety of popular machine learning frameworks, including TensorFlow, Scikit-learn, and PyTorch. They also allow custom integrations with enterprise data sources. Combined with the Wallaroo observability tools, the platform delivers compute and model performance metrics in a format that is easy to understand.
For businesses seeking to scale their AI investments, the Wallaroo ML platform may be the answer. Its observability capabilities include the ability to compare the performance of a model with the help of metrics such as the number of simulated events, the number of predicted outcomes, and the average accuracy of a model. In addition to the aforementioned, Wallaroo also provides a model management tool that supports the most common ML workflows. And as a bonus, the Wallaroo ML platform boasts an efficient execution engine that enables users to react to market changes in real time.
Lastly, Wallaroo provides a simple and efficient deployment capability that can run multiple models on shared infrastructure with little or no additional overhead. This allows users to save money, improve their ROI, and gain visibility into the performance of their machine learning models. The Wallaroo ML platform is built for speed, ease of use, and robust analytics. By leveraging the Wallaroo ML platform, companies can get the most out of their data and achieve the most powerful machine learning results possible.
Investments in product development and hiring
Wallaroo Labs has just received $25 million in Series A funding from Microsoft’s M12 venture arm. The company plans to invest the money in product development, hiring, and sales. The startup’s innovative enterprise AI platform has already been deployed by the US military, the Fortune 500, and other large companies. Using analytics to transform the data that enterprises generate, Wallaroo Labs provides its customers with the value that they need to achieve their goals.
Wallaroo offers a comprehensive, scalable solution for deploying ML models and analytics into production. The platform includes model management, data connectors, observability tools, and audit and performance metrics. These features allow users to easily and quickly evaluate and deploy ML models against live data. They also allow for the easy creation of custom integrations with in-house solutions. In addition, Wallaroo supports a number of popular machine learning frameworks, including TensorFlow, PyTorch, and Scikit-Learn. Its engine can run ML models at the edge or in the cloud, and it compiles to native code. With a single server, Wallaroo can analyze over 100,000 events per second, making it the fastest machine learning platform in the world.
Wallaroo’s observability tools deliver compute, and model performance metrics, as well as audit logs and A/B testing comparisons. These advanced capabilities enable more granular model validation checks, better anomalies detection, and faster drift detection. For example, Wallaroo’s observability tool offers model validation in real time, allowing users to measure model performance against baselines. This ensures that users are informed about model underperformance and system bottlenecks.
The Wallaroo platform is a plug-and-play, cloud-based solution that gives users full visibility into live models. It is built on four key components: MLops, a distributed-processing engine, data connectors, and model management. Each component has its own purpose, but together they form a powerful, unified analytics and visualization platform. As a result, Wallaroo is able to provide a scalable, blazing-fast computing platform for a wide range of use cases. From computer vision to security and fraud prevention, Wallaroo has helped businesses across a wide range of industries get the most out of their data.
Model insights
Wallaroo is an advanced machine learning platform designed for enterprises that want to use ML to impact their bottom line. The company is based in New York and is backed by venture capital firms. It has a vision to provide the easiest way to deploy ML in production. Their use cases include fraud, cybersecurity, security, marketing, pricing, and computer vision. They work with Fortune 500 companies and the US Military.
Wallaroo offers two products: Model Insights and Edge Inference. In Model Insights, users can set parameters to measure model performance against baselines. This helps detect and identify model underperformance faster. Additionally, this technology can monitor data drift. With this technology, data scientists have full visibility into live models, including the observability of key model inputs.
In addition, Wallaroo’s Model Edge data platform is an edge device management service that sends model scores and telemetry metrics to the Model Edge data platform. This enables the management of model insights and model registry to and from the edge devices. Using this service, users can automatically manage and deploy their models to all types of constrained edge devices. By combining these services, Wallaroo provides a powerful way to deploy and monitor ML models in production. Moreover, the platform supports on-premises data aggregation.
Wallaroo’s Model Insights uses a data-drift framework to detect and report data drift. It continuously monitors the distribution of model scores over user-selected time windows. Specifically, it can be used to monitor concept drift in credit card fraud models. The system has the ability to identify a change in model accuracy and alert the user immediately. Moreover, this technology can be extended to several different use cases.