Data-as-a-product: A framework for collaborative data wrangling

on YouTube ↗20m

" summary

A reproducible framework called daapr packages data wrangling into versioned, shareable “data products” and is actively used at Bristol Myers Squibb across exploratory analyses, with some repos showing commits since 2022. It versions code via Git/GitHub, environment via renv, and data via pins, uses derivation functions and optional targets pipelines, currently supports RDS/QS and remote boards, plans parquet support, and reports successful collaborative reuse with reader/writer/owner onboarding roles and documented credential and Git workflows.

AI-written summary · about

" chapters (18)

1. 0:00Reproducible data wrangling
2. 1:09Ad hoc analysis challenges
3. 2:35Data products for analysis
4. 3:36DAAPR framework overview
5. 4:19Versioning everything
6. 5:48DAAPR workflow steps
7. 7:14Derivation code in DAAPR
8. 8:21Data product object structure
9. 9:28Versioning and deploying products
10. 10:32Iterating on data products
11. 11:14Collaboration in repos
12. 12:22Adoption barriers
13. 13:26DAAPR onboarding roles
14. 14:48Proving framework success
15. 15:11Adoption roadblocks
16. 16:55User troubleshooting
17. 17:16DAAPR resources and future
18. 18:41Q&A

" related talks

• 24mMake data central featureAntoni Ivanov · DSC Europe 23
• 20mModernizing the Data Science Toolkit of a 40-year-old Market Research CompanyKeaton Wilson · posit::conf(2024)
• 22mData as a Product: Transforming Analytics for Business SuccessRam G Suri, Neeraj Bholani · XConf India 2024
• 25mUnpacking The Data Product Experience So FarANDY MOTT · Big Data LDN 2022
• 16mProduct Eyes – Seeing Data In A New Light To Build Valuable ProductsLEXY KASSAN · Big Data LDN 2022
• 39mData Mesh: Why, What and HowZhamak Dehghani · Craft Conference 2021
• 24mData Versioning Explained the Open Source WayEinat Orr · Developer Summit 2020
• 30mData Products 101: From Data Chaos to ClarityJean-Georges Perrin · ODSC AI West 2025