Optimizing Machine Learning (ML Ops) in Financial Services with Kubernetes and Cloud Technologies.
summary
Bank moves 450 VMs → Kubernetes and rebuilds ML pipelines with Docker, Terraform, CI/CD and public clouds (AWS/Azure) to cut maintenance and operational toil, reporting about 70% fewer infra incidents and fault events dropping to near zero. Fraud model deployment shifts from old processes to containerized ML and raises detection from 98.9% to 99.96% accuracy. Also adds LLM-style automation for runbook decisions and emphasizes training models on small curated datasets.
AI-written summary · about
related talks
06mEvolving with Kubernetes: Embracing Model Ops
32mMaking Models Work: How an optimized ML lifecycle can drive better ROI
26mMachine Learning Meets K8s
21mMastering the DevOps Machine Learning Pipeline
14mPerformance Optimization in High-Volume Transaction Systems
18mArchitecting Intelligent Platforms
28mMachine Learning on Kubernetes at Shell: A Kubeflow Journey
46mMachine Learning with Kubernetes