Executive Spotlight: Who Is Josh Wills, Member of Technical Staff at DatologyAI?
Josh Wills is a Member of Technical Staff at DatologyAI, a company that curates training data for AI model builders. He is best known as an early data science leader at Cloudera, the founder of Apache Crunch, and the former head of data engineering at Slack.
Per a recent MotherDuck interview, he has been at DatologyAI since 2024. He is an engineer rather than a corporate officer, so this page covers his career and the company he works for.
This page uses public professional information only.
Current role and company
MotherDuck's blog post on its podcast conversation with Wills says he has been a Member of Technical Staff at DatologyAI since 2024. It says he builds tools that select training data for deep learning models, along with the distributed pipelines that process that data. DatologyAI's own homepage does not mention him or publish a team page, so the title rests on that interview.
- TitleMember of Technical Staff, DatologyAI (per MotherDuck).
- Since2024 (per MotherDuck).
- CompanyDatologyAI, which calls itself a data refinery for AI model builders.
- FocusTools and distributed pipelines for selecting training data.
Career path
Wills has spent his career moving between data infrastructure and applied machine learning. The dates below come from the MotherDuck interview and a 2015 ODBMS.org profile; where a source gives no date, we give none.
- Education. ODBMS.org lists a bachelor's degree in mathematics from Duke University and a master's in operations research from the University of Texas at Austin.
- Google, 2007 to 2011. Staff engineer who worked on the ad auction system and then led analytics infrastructure for Google+.
- Cloudera. Led data science. ODBMS.org lists him as Senior Director of Data Science in February 2015, and an InfoQ page for a QCon talk listed him as director of data science in January 2014.
- Slack, 2015 to 2019. Led data engineering.
- 2022 to 2024. He says he had basically retired in 2022, until AI got him back, and he joined DatologyAI in 2024.
What he is known for
- Apache Crunch. MotherDuck says he founded the Apache Crunch project while at Cloudera.
- A book on Spark. He co-authored the O'Reilly book Advanced Analytics with Spark.
- dbt-duckdb. MotherDuck describes him as the original creator of the dbt-duckdb adapter, which connects the dbt transformation tool to DuckDB. It gives no date for his work.
- A conference speaker. He gave a QCon talk titled From The Lab To The Factory: Building A Production Machine Learning Infrastructure, listed by InfoQ on January 16, 2014.
Leadership style and philosophy
In the MotherDuck conversation he says data engineering is fundamentally the cache invalidation problem, applied in a batch-processing context. He also says the thing engineers most need to communicate is the decisions and trade-offs they make.
He argues that AI does best where feedback loops are fast, and that large data pipelines lack them. He says he automates boring work such as flaky tests and database migrations so he can focus on high-leverage decisions. The page was read through a summarizing tool, so we paraphrase rather than quote at length.
Key numbers
Wills is not an officer of a public company, so there are no financial results to report. The figures below are DatologyAI's own claims from its homepage as read on October 31, 2026, and are company-reported rather than independently verified.
- Thomson Reuters case studyMid-training on a legal corpus gave +5% on Legal Bench and +2.5% on general evals, at under 1% of base pre-training budget (100B mid-training tokens vs 15T pre-training). Company-reported.
- Arcee case studyA 398B-parameter mixture-of-experts model (13B active) trained on 17T tokens curated by DatologyAI, built by about 10 people. Company-reported.
- Arcee usage3.37T tokens served on OpenRouter in the first two months. Company-reported.
- Wills at DatologyAISince 2024 (MotherDuck).
What to watch next
Worth tracking: DatologyAI's research posts, such as its October 9, 2026 post on Curation Studio, and Wills's public talks and interviews on data for AI.
Questions people ask
Who is Josh Wills?
He is a Member of Technical Staff at DatologyAI and a former head of data engineering at Slack and data science leader at Cloudera, per MotherDuck and ODBMS.org.
What is Josh Wills's current role?
Member of Technical Staff at DatologyAI, where MotherDuck says he has worked since 2024.
Where did Josh Wills work before DatologyAI?
Google as a staff engineer from 2007 to 2011, Cloudera in data science, and Slack, where he led data engineering from 2015 to 2019.
What is Josh Wills known for?
Founding Apache Crunch, co-authoring Advanced Analytics with Spark, and creating the dbt-duckdb adapter, per MotherDuck.
What does DatologyAI do?
It calls itself a data refinery for AI model builders, curating training data for pre-training and mid-training and not selling data.
Sources
- MotherDuck: Data engineering is the cache invalidation problem ft. Josh Wills · Current role, career timeline, views
- DatologyAI homepage · Company description and case-study figures
- ODBMS.org: Josh Wills, Cloudera (Feb 9, 2015) · Cloudera title, education, Google background, photo
- InfoQ: Machine Learning Infrastructure (QCon) · Talk and 2014 title
Facts were read from the linked pages on October 31, 2026. Some pages were read through a fetch tool that returns summaries, so confirm details on the original page. The hero photo is a 2015 headshot from ODBMS.org, centered on a card. This page covers public professional information only.
Top photo: ODBMS.org (odbms.org/2015/02/josh-wills-cloudera/)