Operationalizing Machine Learning for Developers

For years, the procedure for scoring data (running predictions) against machine learning models has been pretty manual and ad hoc. Like the process for building the models in the first place, data scientists have done the work as a series of one-off tasks on their own computers. But that’s not ready for prime-time. It doesn’t scale, it can’t be scheduled, and it excludes mainstream developers from participating in the AI revolution that the industry claims is taking place. If we want AI to be mainstream, and if we want line of business applications to embed intelligence routinely, we need to make scoring data against a model as simple for developers as calling a function. That means getting the models into mainstream server environments and making them accessible through REST APIs, database stored procedures or local function calls in mainstream programming languages. Luckily, several software companies, including one of the biggest, are doing just that. In this session, Big Data blogger and industry expert Andrew Brust reviews several of the newer options available for operationalizing machine learning models. Andrew will outline the mechanics of how these technologies work, explain which developer constituencies they embrace, and will even demo one or two of them. Come to this session if you want to see how Enterprise development and AI can get together.