The goal is to get machine learning techniques into the hands of industrialists. The challenge is the complexity involved in accurately training machine learning algorithms. This typically requires data science expertise, but as more and more generalists become more and more tech savvy, the power of machine learning can be put in the hands of the folks closer to the data. In the IIOT world (Industrial Internet of Things), Operations Engineers are often the most familiar with the data coming off the sensors of industrial equipment.
As Principal Product Designer, I work with stakeholders (customers, product managers, business leaders) to help define product roadmaps. I work with international scrum teams (California, Bangalore, New York) to ensure consistent user experience, and make sure the tools are intuitive and meet user needs.
Storyboarding and concept flows help stakeholders, product managers, engineering teams, etc. easily grasp the high-level conceptual flow of the product.
I traveled to several customer sites to examine user workflows and gain critical insights into user behaviors. Industry roles for this line of products include Reliability Engineers, Data Analysts, and Data Scientists.
During interviews with subject matter experts, it becomes obvious that data exploration and data visualization is a critical part of the process. Data visualization tools let users evaluate data before and after analytic processing.
The data output of the deployed analytic models feed into prognostic dashboards, giving users insights to make better on-the-job decisions.
The primary personas for these dashboards are Reliability Engineers and Performance Analysts. The visual look for the equipment monitoring dashboards are dark because industrial equipment monitoring facilities are often in lowlight environments.
The data science process is very iterative, and often never complete because machine learning models might get “stale”. I led the team in designing monitoring tools so that Operators can ensure that the predicted results are close to actual results. If not, the machine learning model may have to be retrained.
I have influenced many design patterns across GE Digital. As Lead Product Designer for several software products at GE Digital since 2013, my product design contributions have made a substantial impact on the GE design system.