AI System on Native AWS - Part 8 - 即時串流 ML 與詐欺偵測

詐欺偵測是即時 ML 的極限測試:要在幾十毫秒內對每筆交易做出放行或攔截的決定,特徵要用『此刻及過去幾秒』的行為即時算出,對手還會主動規避你的規則。本篇用純 AWS 原生服務打造即時串流風控:Kinesis 收交易流、Managed Service for Apache Flink 做串流特徵、SageMaker/Fraud Detector 毫秒級評分、Neptune 圖資料庫抓詐欺團夥、DynamoDB 當線上特徵與決策存放,全部用 CDK(CloudFormation)描述,深入談串流特徵一致性、時間窗、圖偵測與規則+ML 混合決策。

·27 min

AI System on Native AWS - Part 3 - 即時個人化推薦系統

推薦系統是最經典、商業價值最直接的 AI 系統。本篇用純 AWS 原生服務打造一套即時個人化推薦:Kinesis 收即時行為、SageMaker Feature Store 管線上/離線特徵、SageMaker Endpoint 做低延遲推論、DynamoDB 當候選集與快取,並用 API Gateway + Lambda 對外服務。全部用 CDK(CloudFormation)描述,深入談 online/offline 特徵一致性、召回+排序兩階段、冷啟動與 A/B 測試。

·26 min

Building a Sentiment-Driven US Stock Trading System with X.com Real-Time Analysis

Learn how to architect a production-ready sentiment-driven stock trading platform that streams X.com posts in real-time, analyzes market sentiment using AWS Bedrock and HuggingFace, and automatically executes trades on US stocks based on social media signals.

·20 min

Building Centralized Logging with OpenSearch and AWS CDK

Learn how to build a scalable centralized logging platform using OpenSearch, Kinesis Data Firehose, and AWS CDK to collect, process, and analyze logs from Kubernetes clusters and containerized applications.

·16 min