ML news

by Netflix Tech

How Netflix Keeps Large GPU Jobs Running Reliably

Reliable GPU compute for large training & inference jobs
...
...

How Netflix Automatically Right-Sizes Media Containers

Netflix Stratum Media Processing: Automated Container Right-Sizing
...
...

Building and Monitoring Trustworthy LLM Judges

The Lifecycle of LLM-as-a-Judge: Building, Aligning, and Monitoring at scale
...
...

How to Tune Spark at Production Scale Without Losing Track

Running Apache Spark experiments in my sleep (and on a plane)
...
...

Netflix Conductor: Scaling Reliable Workflow Orchestration

Netflix Conductor: The Next Chapter
...
...

How Netflix Rebuilt Ads for Live Streaming at Scale

Behind the Scenes: Evolving Netflix’s Ads Event Pipeline for Live — Part II
...
...

How Netflix Built Cost-Efficient Cassandra Cold Storage

TimeSeries Tiered Storage Journey: Kafka/Flink Streams to Native Cassandra Cold Reads
...
...

Measuring the Impact of Personalized Recommendations

Measuring the Impact of Personalized Recommendations
...
...

How Netflix Built a Trusted Metrics Layer with DataJunction

DataJunction as Netflix’s answer to the missing piece of the modern data stack
...
...

How Netflix Uses Foundation Models for Recommendations

Integrating Netflix’s Foundation Model into Personalization applications
...
...

Modern Java: Build with Modules, Not the Class Path

Leave the Class Path in the Rearview Mirror
...
...

MAPS: Netflix’s Multimodal Asset Personalization at Scale

MAPS: Netflix’s Multimodal Asset Personalization at Scale
...
...

How Netflix Scales Flink: From Custom to Open Source

A Tale of Two Flink Autoscalers
...
...

How Data Projects Make Netflix Workflows Resilient

Data Projects: Managing Data Assets at Netflix Scale
...
...

How Netflix Uses Data Canaries to Catch Catalog Failures

The Data Canary: How Netflix Validates Catalog Metadata
...
...

How Netflix Uses Kueue to Scale Batch Jobs Efficiently

How Netflix Simplified Batch Compute with Kueue
...
...

How Vera and VOID Make AI Video Editing More Precise

Toward More Controllable AI Video Editing: An Early Research Exploration at Netflix
...
...

How Netflix Generates Personalized Homepages with AI

GenPage: Towards End-to-End Generative Homepage Construction at Netflix
...
...

Building Real-Time Service Topology at Netflix Scale

Building Service Topology at Scale: Architecture, Challenges, and Lessons Learned
...
...

How Netflix Built Its Production LLM Serving Platform

In-House LLM Serving at Netflix
...
...