Uber

AI-driven mobile testing system using large language models.

Extends Michelangelo for end-to-end LLMOps and prompt engineering.

Anonymizes sensitive information in requests for enhanced security.

Unified platform for accessing LLMs from multiple vendors.

Enhances neural network training with loss change allocation.

Open source Python package for AI unit testing.

Powers food recommendations using graph learning techniques.

Controls text generation using advanced AI language models.

Measures model performance at Uber-scale using AI techniques.

Open-source deep learning toolbox with new integrations.

Open-source platform for building conversational AI agents.

Optimization tool reducing ML model tuning costs significantly.

Novel architecture for deep learning in temporal prediction.

In-house tool for parallel feature search in ML models.

AI-driven system for unsupervised anomaly detection in time series data.

Improves home feed recommendations by reducing bias, enhancing user experience.

Governance system overseeing 120,000 data workflows for enhanced management.

AI-driven deep learning for more accurate Uber arrival predictions.

Optimizes ad placements using AI to predict advertising outcomes.

Identifies, explains fraud anomalies using histograms in Uber's process.

Detects fraudulent activities using unsupervised machine learning and anomaly models.

Enhances recommendation systems with efficient user and store embeddings.

Simplifies AI distributed computing, enhancing model training efficiency and scalability.

Promotes varied learning challenges and solutions in reinforcement learning.

Few-shot link prediction through meta-learning for predictive analytics.