Published May 31, 2021

Buy AND Build for Production Machine Learning with Nir Bar-Lev - #488

Nir Bar-Lev delves into the complexities of federated learning, MLOps evolution, and experiment management, sharing insights on balancing engineering with data science, the buy vs build debate, and managing wide versus deep machine learning platforms. As MLOps grows, he addresses integration challenges and strategic solutions essential for future success.
Episode Highlights
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) logo

Popular Clips

Episode Highlights

  • Buy vs Build

    The debate between buying and building MLOps solutions is nuanced, with emphasizing the importance of integration. He notes that AI involves models, data, and code, making integration complex on multiple levels, including technical and operational 1. Nir suggests a hybrid approach, where companies buy a modular platform and build specific solutions on top of it, allowing flexibility and customization 1. He also highlights that experiment management is increasingly a "buy" decision, as data scientists prefer ready-made tools to focus on their core tasks 2.

       

    MLOps Growth

    The MLOps industry is poised for significant growth, yet it faces challenges in keeping pace with technological advancements. observes that while sophisticated MLOps tools are emerging, many industries lag behind in adopting these technologies 3. He predicts that MLOps will proliferate across various sectors in the next decade, bridging the gap between cutting-edge technology and industry needs 4.

Related Episodes