Artificial Intelligence1 min read

How Meta Uses AI Agents to Scale ML Innovations Across Ads Ranking Models

By · Published by Everything Blog

In short

The automated framework expands on earlier template-based building blocks by deploying autonomous agents capable of making complex judgment calls during technique propagation. These agents currently evaluate candidate architectures, diagnose underperforming models, and resolve runtime operational failures across varying system constraints. Over the past two years, initial template automation reduced per-model engineering time by 92% across 5,000 combinations. The subsequent introduction of autonomous agents has further driven a threefold rise in overall engineering productivity.

Key points

  • Deploying agents to manage propagation decisions allows engineering teams to realize subs…: Deploying agents to manage propagation decisions allows engineering teams to realize substantial infrastructure savings and triple their operational productivity.
  • Proven advancements like Interformer networks and Wukong factorization layers are distrib…: Proven advancements like Interformer networks and Wukong factorization layers are distributed efficiently across distinct model architectures.
  • Unlike rigid automation templates, AI agents possess the judgment needed to evaluate hype…: Unlike rigid automation templates, AI agents possess the judgment needed to evaluate hyperparameter scopes and resolve unexpected execution failures.

Sources checked by our analysis pipeline: 1

- Our previous post in this series introduced the Ranking Engineer Agent (REA), an autonomous system that generates hypotheses, designs experiments, and proves new machine learning (ML) techniques on ads ranking models.

- Now, we’re sharing how we propagate ranking model improvements like architectural advancements and training optimizations efficiently across a large and diverse fleet of ads ranking models.

- This post walks through this process that we call “technique propagation” and how we’re optimizing the process using AI agents.

- Since agents began driving execution in March 2026, we estimate AI agents handle over 60% of all technique propagation. Measured against the same work before agents, early results show 3x greater engineering productivity, a median 4x reduction in training compute across select combinations, and an estimated $45M+ in compute savings by the end of the year.

Meta’s ads recommendation system is powered by a large number of distinct ads ranking models, each serving different product surfaces and advertiser objectives, and each with its own architecture, training data, and serving constraints.

Across this portfolio, ML researchers and engineers are continuously…

Artificial Intelligence