Abstract: For the past 5 years I have managed large teams that provide the critical feedback to Data Scientists and machine learning models to achieve higher and higher levels of confidence from machine learning models over time. The work done by the Humans in the Loop is a critical component for successful data science projects. The Humans in the Loop may start out by sharing simple feedback at the start of a project, but the Humans in the Loop will quickly learn to share concrete feedback that can directly improve the confidence of the model and success of the data science initiative. The biggest benefit from having an experienced team of Data Analysts in the loop is their ability to become domain experts in a variety of areas over time through experience, repetition and research. The humans in the loop are often times aspiring Data Scientists who given the opportunity will grow from sharing simple feedback to the model, to sharing critical feedback to the data scientists, and eventually building future machine learning models in their area of expertise.
This talk will go into depth about what it takes to build, manage, and upskill a team of Data Analysts who provide feedback as the Humans in the Loop to a variety of data science models and initiatives. Some teams design data science projects without a human in the loop, although many of the most successful uses of Data Science leverage teams of Data Analysts to provide critical feedback to machine learning models that are in production and consistently trained on new data. The Data Analysts who perform the human in the loop work are some of the best team members to learn the more advanced techniques in Data Science to directly build their own models that power the future of Data Science initiatives. The talk will also share real examples of how Data Analysts have already started making the transition from their work as the Humans in the Loop to their work building machine learning models that directly leverage humans in the loop to deliver confident output across a variety of Data Science challenges.
Bio: Jonathan Rabinovitz is the Director of Data Operations at Oracle within the AI Applications group. He develops the strategy and frameworks to accelerate data science initiatives with dedicated teams of domain experts and Human in the Loop Project Managers. Previously, Jonathan worked for several startups to expand their ability to create exceptional training sets for machine learning models. Over the past 20 years, Jonathan has worked on a variety of complex data projects to transform hundreds of millions of unstructured data points into high quality alternative data sets.