Development · Web app
the-algorithm
Open-source codebase powering X's recommendation engine and timelines.
This repository contains the source code for the services and recommendation models powering X (formerly Twitter), including the For You timeline, Search, and notifications. It reveals how candidate tweets are sourced, scored, filtered, and ranked using graph algorithms and machine learning models. Developers and marketers use it to study social media ranking mechanics and optimize content reach.
What it's for
- Understand the scoring mechanisms behind X's For You timeline
- Identify algorithmic penalties and engagement boosts for content strategy
- Study large-scale recommendation system architecture and graph processing
Features
- Candidate sourcing pipelines for in-network and out-of-network posts
- Light and heavy ranking machine learning models for tweet scoring
- Visibility filtering logic for trust, safety, and spam downranking
- PageRank-based user reputation scoring (Tweepcred)
- GraphJet-based real-time graph traversal and candidate selection
Pricing
Open source. Free and open source under AGPL-3.0
Details
- Type
- Web app
- Category
- Development
- Topic
- Recommendation Systems
- Open source
- Yes
- Platforms
- Self-hosted
- For
- Machine learning engineers, data scientists, and social media strategists
- Tags
- algorithms, recommendation engine, scala, machine learning, social media, ranking
Why it was saved
- @aidenybai: Replace github.com with repogrep.com in any repo URL to instantly ask questions of that codebase; demo shows Fast/Smart modes, semantic code search with per-query timings, cached repo detection, and cited answers with code snippets, file references and practical takeaways.
Related
The open-source code powering the For You recommendation feed on X.
Unified serverless API to discover and run open and closed AI models
Typed TypeScript SDK for publishing, analytics, and interactions across social platforms
Updated