Prod, Do Stack: Production-Ready Solutions via Data Science Competitions

on YouTube ↗33m

" summary

The talk describes Dbrain’s new infrastructure for running machine-learning competitions where participants submit executable source code. Models are trained on client data and predictions are generated in Dbrain’s cloud, and the best solutions are automatically turned into APIs for production use.

AI-written summary · about

" chapters (7)

1. 0:00Introduction
2. 3:18Standard competition problems
3. 5:04Solution via new platform
4. 11:57Cryptocurrency public contest
5. 13:50Internal developer contests
6. 16:19Open source in-house platform
7. 20:30Q&A

" related talks

• 17mProductionizing Data Science WorkloadsAnand Iragavarapu · PyCon India 2020
• 02mDBTB INT Nemanja SpasojevicNemanja Spasojevic · Data By the Bay 2016
• 03mDBTB INT Anthony Goldbloom rAnthony Goldbloom · Data By the Bay 2016
• 16mDeep Learning in Production Seen by a Data EngineerPar Romain Sagean · DataXDay 2019
• 22mProductive Machine Learning and Deep Learning ProjectsGreg Werner · MLconf Atlanta 2017
• 01h 25m[Evals Workshop] Mastering AI Evaluation: From Playground to ProductionDoug Guthrie, Carlos Esteban · AI Engineer World's Fair 2025
• 39mDataEngBytes 2023 - BRI-T1-02 - Mikiko BazeleyMikiko Bazeley · DataEngBytes Brisbane 2023
• 28mThe Coding Portion of Data ScienceMilos Milovanovic · Data Science Conference 5.0