Comet: MLOps for the Next Generation of Machine Learning
November 18, 2021
We’re excited to welcome Comet into the OpenView family as we lead their $50M Series B.
Backed by thousands of users and multiple Fortune 100 companies, Comet provides a complete MLOps platform that empowers data science and machine learning practitioners to track, monitor, iterate, and collaborate across the full ML lifecycle – from model design and development to experiments and models in production. They’re helping ML teams across the world build better, more accurate AI models, all while improving team productivity and collaboration.
A market in transition
Comet’s co-founders Gideon Mendels and Nimrod Lahav founded Comet in 2017 with a flagship product that solved one of the biggest challenges ML teams were facing at the time: experiment management. Building a model is a highly iterative process with a wide range of variables that could be affected by many different stakeholders at different points in time. They knew teams needed a centralized location to keep track of the changes they were making to their model and compare experiments side-by-side. Multiple companies at the forefront of machine learning adopted this product, like Uber, Etsy, and Ancestry.
Since its founding, enterprise investment in machine learning has continued to grow rapidly, and machine learning is now moving from an early adopter technology with relatively few models in production to a ubiquitous component of the enterprise software stack. As such, the “jobs to be done” for an ML practitioner are also changing. Responsibilities like production monitoring, understanding business impact, resource allocation, and explainability are now taking center stage.
Helping ML teams find success beyond experimentation
Comet’s capabilities have grown in lockstep with this market maturation, and their experimentation management platform has expanded to include features like model monitoring, dataset versioning, hyperparameter optimization, collaborative workspaces, reporting, and more. Extending Comet across multiple stages of the ML lifecycle allows them to bring together knowledge from models’ behavior in training and in production, context which is key to building and maintaining peak performance. Unified experimentation and monitoring capabilities will be critical for the next generation of data scientists and machine learning engineers, and Comet is the only product on the market bringing them together into a single platform.
Today, Comet is used by organizations of all shapes and sizes, from industries at the cutting edge of AI to those taking on major ML initiatives for the very first time. They’ve grown ARR 5X in the last 12 months and have more than tripled the size of their team. We know the best is yet to come for Gideon, Nimrod and the rest of the Comet team as this fast-growing market moves from its early experimental phases to mainstream adoption. We’re excited to support them in their mission to help every enterprise achieve success in machine learning.