Interview
summary
Claims a production platform that turns batch ML into continuous training pipelines and serves models with live streaming updates. Implements end-to-end pieces using Apache Spark and TensorFlow and deploys on Kubernetes and GPUs to run realtime inference and online learning. Emphasizes intelligent scaling to grow or shrink model compute, coordinates a Node.js UI, and surfaces operational best practices learned from Netflix and Databricks experience. Performance and correctness aim to match offline accuracy while updating continuously.
AI-written summary · about
related talks
05mInterview
33mDeploying and Scaling Spark ML and Tensorflow AI Models
28mBuilding Developer-Centric ML Inference Platforms
26mEngineering Excellence: Platforms and Architectures for Real-Time AI Analytics
42mThe Challenges Managing a Kubernetes-Based Machine Learning Infrastructure
02mInterview
42mDeep learning beyond the learning
39mOperating Deep Learning Pipelines Anywhere Using Kubeflow