This Week in ML & AI avatar

Dexa/This Week in ML & AI

Learn more
Published Oct 24, 2022

You're not Facebook. Architecting MLOps for B2B Use Cases with Jacopo Tagliabue - #596

Jacopo Tagliabue delves into architecting MLOps for B2B companies, offering practical insights on rapid experimentation, strategic build vs. buy decisions, and scaling frameworks to suit smaller enterprises, emphasizing the need for simplicity and flexibility to meet company-specific needs.
Episode Highlights
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) logo

Topics covered

Popular Clips

Episode Highlights

  • Scaling MLOps

    discusses the challenges of adapting MLOps frameworks for smaller B2B applications, emphasizing the need for pragmatic approaches. He highlights the gap between large-scale MLOps practices, like those at Uber or Pinterest, and the more modest needs of smaller companies. Jacopo notes that most businesses fall between these extremes, requiring tailored solutions that balance data, modeling, and tooling 1.

    Most people are actually in the middle of this, you know, this distribution.

    ---

    He stresses that B2B companies face unique challenges in scaling their models, as they often deal with multiple enterprises rather than a single large-scale platform 2.

       

    B2B Challenges

    B2B companies encounter distinct challenges in adopting MLOps, particularly concerning data and tooling. explains that B2B firms often have less data, which complicates model deployment and scalability 3. Additionally, the lack of direct control over data sources makes standardization difficult, as companies must rely on client cooperation for data collection and integration 4.

    It's one of those problems that cannot be solved just by having good code.

    ---

    Despite these hurdles, Jacopo sees significant opportunities for B2B companies to leverage ML, provided they navigate these complexities effectively 5.

       

    Team Structures

    Effective team structures are crucial for B2B companies to maximize their MLOps capabilities. advocates for end-to-end ML roles, where individuals handle everything from data extraction to model deployment, fostering ownership and efficiency 6. This approach reduces the need for large teams, allowing smaller companies to operate effectively with fewer resources.

    One more expensive person will actually cost you way less than five cheap ones.

    ---

    He also emphasizes the importance of building a strong community presence to enhance hiring practices and attract talent 7.

Related Episodes