X Algorithm is an open-source recommendation system for social media platforms that ranks and delivers content to user timelines based on real-time engagement signals.

• Utilizes a 48-million parameter neural network (Heavy Ranker) to predict user interactions including likes, retweets, and replies.
• Aggregates content candidates through SimClusters and GraphJet to identify relevant in-network and out-of-network tweets.
• Implements visibility filtering and safety heuristics to exclude prohibited content and maintain compliance with platform policies.
• Leverages the Twisearch engine and Thrift protocols for cross-language communication and data retrieval.
• Integrates with TensorFlow for machine learning model training and Bazel for build automation.

Ideal for researchers studying large-scale recommendation architectures, developers auditing algorithmic bias, and engineers building personalized content delivery pipelines.