Tecton is a San Francisco-based startup cofounded by Mike Del Balso, Jeremy Hermann, and Kevin Stumpf, who previously worked together at ride-hailing and transportation company Uber, where they co-developed Uber's Michelangelo machine learning platform.
The company was founded in late 2018 and formally launched in late April 2020. As part of its public launch, the company also announced that it had raised $25 million in combined seed and Series A funding co-led by Sequoia Capital and Andreessen Horowitz. Other institutional investors in the company include SV Angel, Lux Capital Management, Liquid 2 Ventures, and Fathom Capital. Angel investors in the company include Mesosphere founder Florian Liebert, Applied Intuition founder Qasar Younis, and venture investor Bucky Moore.
Tecton is a machine learning platform. The company says it expands on the feature store architecture the founding team helped build at Uber. Tecton manages the end-to-end lifecycle of features for machine learning systems that run in production.
The company states its platform consists of the following:
- Feature Pipelines for transforming raw data into features or labels
- A Feature Store for storing historical features and labeling data
- A Feature Server for serving the latest feature values in production
- An SDK for retrieving training data and manipulating feature pipelines
- A Web UI for managing and tracking features, labels, and data sets
- A Monitoring Engine for detecting data quality or drift issues and alerting
The company says that its platform is designed to help machine learning teams with a number of key tasks. These include developing standardized features, labels, and data sets for machine learning from batch and real-time data. Tecton says its platform helps with deploying and serving feature data to models in high-scale production environments. The platform can help teams curate and govern an organization-wide repository of standardized, production-ready features for sharing and reuse across its projects. Additionally, the company says Tecton helps monitor feature data to ensure that models continue to operate correctly over time.
The company's three cofounders met while working at Uber, where they developed Michelangelo and worked with many internal teams on a variety of business and engineering challenges.
- Mike Del Balso is cofounder and CEO of Tecton. Prior to Tecton, Del Balso was a senior product manager of Uber's machine learning platform infrastructure and applied machine learning teams from November 2015 through 2018, when he left to start Tecton. Del Balso led the development of Uber's AI strategy, reporting to the CEO. Prior to Uber, Del Balso was a product manager at Google, where he oversaw machine learning teams working on the company's ads platform.
- Kevin Stumpf is a cofounder and CTO of Tecton. Before starting Tecton in 2018, Stumpf served in a variety of management and leadership roles at Uber, where he started as a senior software engineer and ended as an engineering manager on the company's machine learning platform team, which built Michelangelo and other internal infrastructure and tooling. Prior to Uber, Stumpf cofounded and was CTO of Dispatcher, Inc., a trucking and logistics dispatch platform described as "Uber for long-haul trucking." Since May 2020, he is also a member of Sequoia Capital's scout program.
- Jeremy Hermann is cofounder and VP of engineering at Tecton. Before starting Tecton in 2018, Hermann was Uber's head of data infrastructure from 2014 through 2015. In 2015, Hermann started and led Uber's Michelangelo machine learning platform team, which additionally built infrastructure and tools for engineers and data scientists at the company. Prior to Uber, Hermann served in VP and director-level engineering roles at Hearsay Systems, ZangZing, and Vontu (which was acquired by Symantec).
Tecton: Machine Learning Platform from Uber with Kevin Stumpf - Software Engineering Daily
June 3, 2020