Superlinked's CEO Daniel Svanova and COO Ben Gutkovich engaged me to write a case study on the application of their VectorOps infrastructure to improve feed personalization outcomes for Entre, a networking platform for tech professionals. Daniel provided some background context material and a cursory brief of how Superlinked used semantically embedding to add relevancy (to popularity and recency) to Entre's feed hierarchy, permitting fine-tuning, A/B testing, and KPI optimization - even with very little user and interaction data.

Approaching a topic I knew little about, I needed to do some substantial background study. I also realized I would need some back and forth with one or more Superlinked team members to fill in the blanks of the Superlinked vectorization process and the specific details of their process with Entre. This was not a simple editing job. It required an extensive question-answer process to extract essential data. Finally, I needed to set some early milestones for drafts to ensure our working document was heading towards a final product that would do a superlative job of not only explaining Superlinked's technology and process, but defining a direction for future development.

After reoutlining Daniel's brief, and doing an intensive but quick background read on vectorization and semantic embedding, I began asking several clarifying questions. Once it became obvious that the CEO had limited time, I asked for additional input from team members who might be able to fill crucial gaps. Daniel put me in touch with Superlinked's Tech Lead, Balázs Kemenes. Over the next two weeks, Balázs and I collaborated not only on articulating the details of Superlinked's work with Entre but also opened up discussions with the CEO and COO on both strategy and future general development and documentation directions. In the course of our discussions, I radically and iteratively revised the outline, and assisted Superlinked in articulating their product and process.

Technical marketing document.



Ongoing collaboration with Superlinked on internal documentation and future case studies.