Data Symphony: Orchestrating Insights in Fintech
summary
PayPal handles ~400M active accounts and $1.5T annual payment volume; the project consolidates disparate fraud signals into a centralized fraud indicators dashboard so decision teams stop stitching across teams. Implementation uses notebook + SQL queries to join multiple databases, basic visualizations and KPIs. Initial models took too long and would only refresh every 3 days; options were shorten a 3‑year lookback, accept 3‑day refresh, or fix performance. Chosen path: postpone, optimize runtime, ship a daily MVP that users adopt, then scale.
AI-written summary · about
chapters (9)
- 0:00Introduction and agenda
- 1:15PayPal’s data scale in fintech
- 3:05Role of a data science PM
- 6:42Fraud prevention use case
- 12:56Three key takeaways
- 13:18Know your data deeply
- 15:06Validate hypotheses with data
- 16:45Tell the data’s story
- 18:58Summary and closing
related talks
31mData Science for Financial Inclusion
25mAnalyzing Daily Bank Transactions: Maximizing Insights
27mMaking Money with Data Science
01h 19mOn financial modelling, credit scoring, and risk assessment
49mUnderstanding Customer Journeys with Deep Learning
36mML: Predictive analytics in Ad Engineering
03mHighlights
22mReal-world applications of machine learning: Beyond the hype