You’ve gathered your data, built your machine learning model, linked libraries and code. Now comes the hardest part: Deploying your machine learning model. Pure code deployments have benefited greatly from automated CI/CD pipelines that have made continuous integration and continuous deployment practically seamless. But deploying machine learning models doesn’t yet have that luxury–and deployment remains one of the most painful parts of the entire process. How do we make ML deployments painless? When so many models need lots of libraries and code mixed into a long and unwieldy chain, it can seem impossible to make inference simple–or at least more seamless. In this program, we’ll talk with some top AI/ML experts about how to simplify the process of deploying ML models and serving them to powerful end-user applications.

Key Takeaways:

• Why is deployment so challenging? Is there a one-size-fits-all deployment solution?
• Can we emulate what CI/CD does for pure code systems in ML deployments? What would that pipeline look like? 
• Are organizations better off writing their own deployment engines?
• Are there any emerging standards in deployment?