MLOps Week 29: Building the Future of ML Platforms with Ketan Umare
May 30, 2024
auto_awesome
In this podcast, Ketan Umare, CEO of Union and former leader of the ETA models team at Lyft, discusses the evolution of ML platforms. Topics include the importance of collaboration in AI, challenges of using Airflow in ML operations, building ML platforms with correctness and collaboration, and the future of AI workflows. Insights on navigating the changing landscape of AI and building scalable infrastructures are shared.
Building ML platforms requires collaboration among AI, data, and software engineers for successful product delivery.
The future of AI includes a blend of ML and LLM tools alongside traditional models like random forests.
Deep dives
The Creation of Flight Platform
The podcast delves into the backstory of the Flight platform, detailing how it originated from challenges faced at Lyft in creating accurate ETA predictions for users. K-thon, the CEO at Union AI and Lyft alumnus, shared his experience of using machine learning to optimize these predictions, highlighting the complex factors like traffic and user behavior that influence accuracy. This led to the early stages of building models within Lyft, which eventually evolved into the need for a scalable and operationalized machine learning platform.
Airflow's Limitations and Need for a New Solution
The episode explores the limitations of Apache Airflow in handling complex machine learning pipelines, emphasizing the challenges when deploying models in real-time scenarios. K-thon discusses issues such as version conflicts, lack of experimentation capabilities, and the need for rapid local experimentation before production deployment. He points out the importance of semantic correctness and the need for portable and cost-effective solutions to run and ship models efficiently.
Evolution of AI Products and Organizational Collaboration
The conversation shifts to the evolving landscape of AI products and the necessity for collaboration among data engineers, ML engineers, and software engineers for successful product delivery. K-thon highlights the constant evolution and retraining required for AI models, contrasting with traditional software products. He stresses the challenges posed by organizational silos and the importance of cohesive teamwork between different engineering disciplines.
Future of ML and AI Integration
The podcast concludes with a discussion on the future of machine learning and AI integration, suggesting that AI disruption primarily affects legacy software systems. K-thon predicts that traditional machine learning models like random forests will remain prevalent in production applications due to their ease of use and cost-effectiveness. He anticipates a blend of ML and LLM tools in various applications while underscoring the ongoing need for statistical understanding and efficient data management in AI development.
Join Featureform’s Founder and CEO, Simba Khadder, and Union CEO and co-founder, Ketan Umare, as they delve into Ketan’s journey, starting with leading the ETA models team at Lyft, the origins and evolution of Flyte, an ML workflow platform, and his latest venture, Union. The discussion also covers the importance of collaboration in AI, the future of traditional machine learning in the era of LLMs, and the potential disruptions in the software industry. Whether you're a data scientist, engineer, or AI enthusiast, this episode offers valuable perspectives on building scalable ML infrastructures and navigating the rapidly changing landscape of artificial intelligence.
Get the Snipd podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share & Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode