Building a Data-Driven Bank
summary
Case study of a bank that builds a centralized data platform using HDFS/S3, Kafka, Databricks, MLflow and Kubernetes plus in-house tools (Cobrix to parse mainframe COBOL files, Py2K/Abris/Hyperdrive/Subatomic, and feature‑ and model‑maker). Cobrix moves mainframe data into Spark, cutting processing cost to a small fraction and avoiding vendor lock‑in. First model takes ~9 months to deliver; subsequent models deploy in 2–4 weeks or ~1 month. Leads response jumps ~300% and product take‑up improves 2–15x, so the approach works.
AI-written summary · about
chapters (10)
- 0:00🚀 Build Stuff Conference – Get Your Tickets!
- 2:00Introduction to Data-Driven Banking
- 5:00The Opportunities in Banking and Other Industries
- 11:20Challenges in Data Access and Discoverability
- 15:16Overcoming Technical Challenges in Legacy Systems
- 19:37The Data Engineering Approach to Build a Data-Driven Bank
- 24:33Data Platforms: Batch and Real-Time Processing
- 30:29Feature Generation and Machine Learning in Banking
- 36:54The Importance of Bridging Business and Data Teams
- 42:17Future Challenges in Building a Data-Driven Bank
related talks
40mProductionizing Big Data - Stories from the Trenches
26mBuilding a data lake at a bank #bbuzz
35mHigh-performance data foundation for AI success
45mModern Data Architecture
19mLessons from Building a Large-scale, Multi-cloud Data Platform at Databricks
37mBig Data Analytics, Machine Learning, Real Time Processing
30mArchitecting Applications With Multiple Open-Source Big Data Technologies
32mMaking Models Work: How an optimized ML lifecycle can drive better ROI