Best practices for advertising and marketing technology on AWS
summary
Advice: run real experiments not slides. Swap MapReduce text to ORC and cut EC2 by 69%, isolate Spark in containers on EKS and use Lambda for dispatch with throttles and heartbeats, and containerized Spark yields ~55% compute savings. Real-time bidding at 800 billion transactions/day (30M/sec) with <20ms latency favors trying Graviton2 and aggressive Spot + stateless refactors. For ML pick a framework, use SageMaker Autopilot and Feature Store to cut ML resource use ~60% and cost ~50%.
AI-written summary · about
chapters (5)
- 0:00Intro
- 0:38Measurement and analytics
- 6:22Real-time advertising
- 12:18Customer data and identity
- 17:31Innovation with machine learning
related talks
45mStart your AI/ML Projects Right
32mMaking Models Work: How an optimized ML lifecycle can drive better ROI
23mKeynote: Making IoT Experimentation Easy
21mRun Less Software
15mFireside Chat with AWS and Redmonk | AWS Developer Day 2025
38mFantastic ML apps and how to build them
38mData Secrets from a Platform Engineer
49mHow to get smarter while doing your job