Overview
In my first role at Pluralsight I worked on the Recommendations team focused on building an ML based recommendations engine to help learners on the Pluralsight platform close their technology skills gap.
On this team I was the sole designer and worked alongside a Product Manager, Machine Learning Engineers, Data Scientists and Front-end engineers. Our team was at the forefront of building the first machine learning models at Pluralsight. I worked on this team for 6 months exploring how recommendations on the platform could be used to enhance the learners journey. However, due to a company-wide reorganization I moved onto the Assessments team at Pluralsight.
In my time working on this team I accomplished the following:
- Led a major discovery effort with the PM to understand how learners were interacting with our recommendations and identified opportunities for bringing recommendations closer to the learners workflow. I made a research plan for the discovery; conducted both primary and secondary research; facilitated interviews with learners; conducted concept testing; developed interactive prototypes for wizard-of-oz testing, synthesized findings into an opportunity map with my PM and Engineering partners. Based on our discovery we honed in on a key need for users: guidance on what to learn next which led to us experimenting with different recommendation patterns such as embedding recommendations at the conclusion of a course to meet learners in their moment of need.
- Designed an explicit feedback mechanism on recommendations so learners could hide recommendations they were not interested in.
- Personalized the ‘New courses’ carousel to show new courses based on a users interest as well as followed authors. This was based on findings from our research that showed that users often monitored Pluralsight for new content in areas of interest as well as from authors of interest.
- Experimented with ways of providing targeted recommendations to learners based on different user behaviors using multi-armed bandits. This experiment led to a ~50% higher breakthroughs on homepage recommendations. These bandits would later be used and improved by other teams at Pluralsight such as Search.Designed and executed several A/B experiments against various business hypotheses that improved user engagement with content recommendations.
- Delivered a talk on Machine Learning UX to peers at Pluralsight.