Explore the challenges of scalability in machine learning with Dr. Waleed Kadous, Head of Engineering at Anyscale, in this insightful podcast episode. Delve into topics such as large-scale computing power requirements, the significance of attention-based models, and the balance between big and small data. Learn about Anyscale's efforts to address these challenges through their open-source project Ray, a popular scalable AI platform. Gain valuable insights from Kadous' extensive experience at companies like Uber and Google, where he led system architecture and pioneered location and sensing technologies. Discover the latest trends in deep learning, infrastructure challenges in MLOps, and the potential of AI-assisted applications. The discussion also covers the upcoming Ray Summit and career opportunities at Anyscale, making it a must-listen for professionals and enthusiasts in the field of machine learning and AI scalability.
ML Scalability Challenges in Machine Learning - MLOps Coffee Session
Overview
Syllabus
[] Waleed's preferred coffee
[] Takeaways
[] Waleed's background
[] Nvidia investment with Rey
[] Deep Learning use cases
[] Infrastructure challenges
[] MLOps level of maturity
[] Scale overloading
[] Large Language Models
[] Balance between fine-tuning forces prompts engineering
[] Deep Learning movement
[] Open-source models have enough resources
[] Ray
[] Value add for any scale from Ray
[] "Big data is dead" reconciliation
[] Causality in Deep Learning
[] AI-assisted Apps
[] Ray Summit is coming up in September!
[] Anyscale is hiring!
[] Wrap up
Taught by
MLOps.community
Related Courses
-
MLOps | Machine Learning Operations
-
Scalable Evaluation and Serving of Open Source LLMs
-
MLOps at the Crossroads - Challenges and Future Directions
-
From Data Engineering to MLOps: Building a Scalable ML Platform - Reverb's Evolution
-
Designing ML Infrastructure for ML and LLM Use Cases
-
The Long Tail of ML Deployment - Challenges and Solutions
