Olcay Cirit
Senior Staff Research Scientist and Tech Lead at Uber AI
Summary
Work
Education
Writing
DeepETA: How Uber Predicts Arrival Times Using Deep Learning
February 10, 2022Describes Uber's low-latency deep neural network architecture for global ETA prediction, improving accuracy over traditional routing engines by refining ETA predictions using ML models and historical data with real-time signals.
Optimal Feature Discovery: Better, Leaner Machine Learning Models Through Information Theory
May 6, 2021Details an approach to optimize feature evaluation and selection in Uber's feature store using information theory to find compact and diverse subsets of relevant features, addressing issues of feature sprawl and redundancy.
Michelangelo PyML: Introducing Uber's Platform for Rapid Python ML Model Development
October 23, 2018Introduces Michelangelo PyML, a platform at Uber that enables rapid Python ML model development by providing flexibility for data scientists to prototype and validate models, extending the capabilities of Uber's main Michelangelo platform.
Consumer Profiling Using Fuzzy Query and Social Network Techniques
January 1, 2005Proposes a method for consumer profiling based on social network theory and the BISC Decision Support System, exploring the use of social connectivity information for targeted advertisements.
Facilitating direct rider-driver pairing
A patent related to methods for facilitating direct rider-driver pairing, particularly relevant for transportation network systems.
Systems and methods for monitoring and evaluating individual performance
A patent related to systems and methods designed for the monitoring and evaluation of individual performance.