February 21, 2019
The future is intelligent. From image-recognition software used to instantly diagnose cancer in patients to voice assistants learning to understand our every question, product experiences in the future will be customized, contextually-aware, and purposeful — in short, they will be intelligent. Driving this transition of product experiences from the static to the sentient are models; specifically, statistical or machine learning (ML) models that have learned rich, complex patterns from data and can be invoked to make real-time decisions.
In this intelligent future, models are the new code. They will routinely implement functionality that previously needed to be hand-written (e.g., database indexing) and serve as the foundations of key product features (e.g., searching video content). In fact, ML models already generate millions of dollars in revenue and, in a short span, are likely to become foundational to many sectors including finance, security, and healthcare.
However, while the tools to develop production-ready code are well-developed, scalable, and robust, the tools and processes to develop ML models are nascent and brittle. Between the difficulty of managing model versions (propensity_XGBoost_num_trees_50_max_depth_4_Final2.gz, anyone?), rewriting research models for production, and streamlining data ingest, the development and deployment of production-ready models is a massive battle for small and large companies alike.
At Verta.AI, we are building software in the service of models and the teams that develop them. With our tools, we seek to empower data science and machine learning teams to rapidly develop and deploy production-ready models, thereby enabling efficient integration of ML into diverse products.
We are starting on this journey by addressing the critical problem of model management. As data scientists begin to routinely train hundreds to thousands of models, model creation far outpaces model versioning, deployment, and management. The Verta platform seeks to fix this issue. We build upon our research at MIT CSAIL on ModelDB — an open-source model management system used at Fortune 500 companies. We extend ModelDB to support model deployment, analysis, and collaboration, enabling data scientists to manage models across their lifecycle. (Sign up here for an early preview.)
Model management is just the beginning. As we begin to streamline every part of the ML journey, we’re thrilled to have partnered with General Catalyst, Village Global, Unusual Ventures, and incredible angel investors.
If the mission to enable businesses to create intelligent experiences inspires you, join us. We’re currently hiring software engineers, data scientists, and business development leaders in Palo Alto.
The intelligent future beckons.