Modern Analytics in the Cloud-A case for fraud detection
summary
Claim: an event‑driven, serverless pipeline on AWS detects e‑commerce fraud with real‑time detection using EventBridge, Lambda, S3, DynamoDB and either SageMaker/EMR or AWS Fraud Detector. Minimum training data is 10,000 records / 400 frauds; tested model hits AUC 0.84 on 70,000 transactions and reduces fraud rate by ~3% in the first 30 days. Known limits: Lambda 15‑minute timeout, 200 preds/sec API throttling, and cold start/label‑maturity needs.
AI-written summary · about
chapters (8)
- 0:00Modern cloud analytics
- 2:53Monolith to modern data
- 7:49Fraud detection case
- 11:38Rule-based fraud pipelines
- 13:24Benefits and challenges of cloud fraud detection
- 16:10AWS fraud detector setup
- 21:02Serverless fraud architecture
- 24:55Scaling fraud detection
related talks
47mBuilding Near Real-time Analytics Solution on AWS
25mBest practices for advertising and marketing technology on AWS
27mArchitecting Sub-Second Analytics Pipelines
08mReal-Time Data Systems: Speed, Scale, and Resilience
32mGoing Server less for Web Services
44mElastic Stack: The analyst's secret weapon in the fight against insurance fraud
22mBatching Was Yesterday: Real-Time Tracking & Analysis For 100+ Million Visitors
28mAccelerate with confidence through AWS Distributed Load Testing