Explore the evolution of Uber's Michelangelo platform in this 36-minute podcast episode from MLOps.community. Delve into the strategic AI overhaul at Uber, tracing the platform's development through three major phases. Learn how Michelangelo progressed from basic ML predictions to sophisticated deep learning and generative AI capabilities. Discover the challenges faced in Michelangelo 1.0 and the subsequent improvements in versions 2.0 and 3.0, including support for PyTorch, enhanced model training, and integration of cutting-edge technologies. Gain insights into advanced features like the Genai gateway, compliance guardrails, and model performance monitoring systems that streamline and secure AI operations at Uber. Understand the importance of ML education within the company and the debates surrounding legacy system maintenance versus innovation.
Uber's Michelangelo: Strategic AI Overhaul and Impact - MLOps Podcast
Overview
Syllabus
[] Uber's Michelangelo platform evolution analyzed in podcast
[ - 4:50] Weights & Biases Ad
[] Uber creates Michelangelo to streamline machine learning
[] Michelangelo platform's tech and flexible system
[] Uber Michelangelo platform adapted for deep learning
[] Uber invests in ML training for employees
[] Explanation of blog content, ML quality metrics
[] Michelangelo 2.0 prioritizes serving latency and Kubernetes
[] GenAI gateway manages model routing and costs
[] ML platform evolution, legacy systems, and maintenance
[] Team debates maintaining outdated tool or moving on
[] Please like, share, leave a feedback, and subscribe to our MLOps channels!
[] Wrap up
Taught by
MLOps.community
