Expert Answer
Anonymous
Situation: When I joined Proofpoint, I was my group's first ML hire. I was given a high level charter. To bring valuable AI powered features to our product that drives revenue. This was a challenging problem yet it gave me an opportunity to lead and build a team from scratch.
Task: My task was to identify features with a great ROI and drive their execution and deploy them into production.
Actions: The first thing i did was take steps to understand the security landscape and customer pain points. I strongly believe in applying solutions to problems and not vice versa. I set up meetings with PMs and TMEs and through them got involved with some customers too and understood what their biggest shortcomings with the platform were. Next I identified areas where ML and AI could solve those problems and came up with a presentation summarizing my findings. I continuously documented all of this in confluence pages. Through constant feedback from management and stakeholders and I refined the problem definition. Next I came up with solutions for each of the problem areas and sorted the features with the feature with the highest return on investment at the top. I worked on prototypes for these solutions using augmented data and also started building a team from our broader group to help me with these implementations. While this was happening I also created epics and user stories and estimated the effort needed to get the highest priority features into production. Based on this estimate I started hiring ML engineers to help us deploy these into production with a target roadmap of two years. Where needed i also contributed with the implementation. Throughout the process I held an open line of communication and kept all the stakeholders informed of the progress. Worth noting that a couple of my proposals also won the internal hackathon and we also ended up submitting patents for the same.
Result: As a result of this structured and balanced approach we currently have a dedicated ML team with 3 features already deployed on production and several more in the roadmap. We also ended up building a MLOps infra and an evaluation framework to support our roadmap. All of this resulted in approximately 15-20 million USD projected growth for our BU through license upselling and new customer sign ups.