We are ex-Amazonians who owned large-scale data products. We worked with dozens of teams and companies who rely on data to make business decisions and power customer-facing systems. Bad data leads to wasted resources from failing database queries, useless recommendations of machine learning models, and wrong business decisions derived from incorrect dashboards.
While at Amazon, we already worked on data quality tools and developed a tool for defining unit tests for data (Deequ on GitHub). The hard problem of how to define good tests for your query workload remained.
Our mission for Deekard is to provide data quality feedback when you need it. Deekard uses SQL queries to understand how you use a dataset to determine the data checks that are most relevant.
Dustin is co-founder of Deekard. Dustin has over 8 years of industry experience at Amazon as Machine Learning Scientist and Applied Science Manager. He founded a team that built tools for data quality measurement and cleaning, including Deequ. Dustin later owned data products powering the Amazon product search page, used by millions of users every day. Dustin holds a PhD in database systems from Hasso Plattner Institute, Potsdam, Germany.
Tammo is co-founder of Deekard. At Amazon, he led a team of Scientists and Engineers, developing algorithms that improve the product search experience. Prior to that, Tammo worked as Applied Scientists focusing on Machine Learning for data quality. There, he experienced first-hand the importance of data observability for developing and running machine learning based products. Tammo has a research background in Medical Physics and holds a PhD in Statistical Machine Learning from the University of Oxford.