Panel: Kubernetes and GPU Trends for AI in Financial Services
Panelat Data on Kubernetes Day NA 2024 · on Tuesday, 2024-11-12· at 11:15· Salt Palace | Level 2 | 250 A·
summary
Banks report data as a growing gravity into zettabyte-scale projections and use satellite and parking data as alternative signals; infrastructure decisions hinge on SLA proximity, data residency, resiliency and cost (example: trading systems require being colocated to meet 12,000 transactions/sec and millisecond latency). Kubernetes is the de facto runtime for scaling and observability, GPU capacity faces a supply constraint so teams optimize with hybrid CPU/GPU deployment, synthetic and small language models accelerate testing and fine‑tuning while a rough 60/40 training vs inference split guides compute planning.
AI-written summary · about
chapters (9)
- 0:00Panelist introduction
- 1:03Data scale challenges
- 5:38Alternative data implications
- 11:03Kubernetes standard in finance
- 14:18Deployment strategies
- 19:35AI adoption trends
- 29:13GPU optimization for AI
- 32:03Q&A: Synthetic data training
- 37:04Q&A: Kubernetes legacy
related talks
27mHarnessing Kubernetes for Scalable AI/ML Workloads
25mExtending Kubernetes for AI | Lessons Learned From Platform Engineering
41mPanel: AI Infra Best Practices: Enterprise Do’s and Don’ts
16mAI + Kubernetes: What Beginners...
38mBeyond Kubernetes: Adapting To Spe...
32mScaling Kubernetes Clusters for Generative Models: Managing GPU Resources for AI App...
31mThe Future of Data on Kubernetes: From Database Management To AI Foundation
36mProduction GPU Cluster with K8s for AI and DL Workloads