UbiOps

UbiOps, formally known as Dutch Analytics, was established in December 2016 by co-founders Yannick Maltha and Victor Pereboom. The company originated from a data science case study undertaken by students from the Delft University of Technology. Founder and CEO, Yannick Maltha, holds both a Bachelor's and a Master's degree in Technology, Policy and Management, and Systems Engineering, Policy Analysis and Management, respectively, from Delft. His academic background and leadership roles in student entrepreneurial societies spurred his interest in data science and entrepreneurship, leading to the startup's creation. The initial spark for the company, then named Dutch Analytics, came from a project with BAM, a large Dutch construction company, to predict railway switch malfunctions. This early success highlighted a broader market need: a scalable platform to deploy and manage data science models. A significant milestone was securing a €2 million seed investment in December 2019, led by Global Founders Capital and backed by BAM, to expand the team and further develop the UbiOps platform.

UbiOps provides a specialized MLOps platform designed to address the common challenge of operationalizing machine learning models. The firm targets data scientists, AI teams, and IT departments across various sectors, including public services, healthcare, and critical infrastructure, who struggle with the engineering complexities of deploying AI. The core of the business is a software platform that automates the containerization, deployment, and scaling of AI/ML models written in Python or R, turning them into robust microservices with API endpoints. This eliminates the need for clients to manage complex infrastructure like Kubernetes, thereby accelerating the time-to-market for AI applications. The business model is subscription-based, offering different tiers, including a free tier, a pay-per-use SaaS model, and an enterprise-level plan for private cloud or on-premise installations.

The UbiOps platform functions as an AI serving and orchestration layer that can be deployed on any cloud (public, private, hybrid) or on-premise hardware. Key features include workload orchestration with automatic scaling, a model and version management catalog, a workflow builder for creating modular pipelines, and comprehensive monitoring and governance tools. It allows users to connect their own compute environments while UbiOps handles the orchestration, ensuring efficient use of resources like GPUs and CPUs. This approach provides flexibility, helps control costs, and avoids vendor lock-in. The platform is designed for a range of applications, from traditional data science models to generative AI and computer vision, serving clients such as Bayer and BAM. By simplifying the deployment lifecycle, UbiOps enables organizations to run their AI workloads securely and efficiently, bridging the gap between data science development and IT operations.

Keywords: MLOps, AI deployment, model serving, machine learning orchestration, hybrid cloud, private AI, data science platform, AI infrastructure, model management, serverless AI