Accelerating AI with Hyperscaler Platforms
summary
Telekom Serbia migrates its analytics stack from IBM Cloud Era to AWS, using AWS SageMaker, EC2 and Lambda to handle a 300 TB monthly signaling feed and train models that create 10–20 personas from 2 million postpaid customers. The team already built 20 analytical models, automates retraining and dynamic scaling, and exposes persona outputs to campaigns, sales and care. Key tradeoffs are visible: clear cost accounting, cross‑cloud joins, consent and data governance, and bespoke monitoring because native AWS observability didn’t meet needs.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:11Telecom Serbia’s AI journey
- 3:20AI projects structure
- 8:16Core team roles
- 12:34On-premise to AWS migration
- 15:00Network signalization analysis
- 19:00Data engineer’s cloud role
- 23:19Data scientist in AWS
- 26:36Lessons from migration
- 30:13Q&A
related talks
36mNew trends in data modernization: Catalysts for AI/ML analytics on AWS
29mData foundation for analytics and AI
50mTake AI/ML from theory to practice with Intel technologies
27mEnterprise AI at Scale
22mFireside Chat with AWS
31mRealizing Industrial AI at Scale with
48mAccelerate science, research, and innovation with AWS
19mDrive transformation through machine learning