December 2022 Summaries

Ravi Trivedi As an Engineering Manager at Tecton, Ravi leads Product Engineering teams that focus on core user-facing parts of the product, including a powerful feature engineering framework for ML engineers and data scientists. Ravi joined Tecton from Google after 6 years with a background in logging infrastructure for Google Search, where he valued working on smaller projects with closer ties to the product and business. He was drawn to startups and eventually met the founders of Tecton at Sequoia Capital's San Francisco office, which led him to join the company as an early-stage startup. Ravi has found that joining a startup requires rewiring his brain to be more hands-on and proactive in problem-solving and process creation. He is a fan of Excalidraw, a drawing tool that helps him reason about concepts, and credits it with improving his ability to articulate ideas to others. Ravi also values the Tecton company culture, particularly its emphasis on being an "owner" rather than just a "renter," which he believes fosters a sense of ownership and responsibility among team members. In his free time, Ravi enjoys exploring new ski resorts in Switzerland, which he considers close to heaven on earth.

Online recommendation systems generate recommendations for users based on real-time contextual information, such as user behaviors and item catalog data. These systems consist of several components, including candidate generation, feature retrieval, filtering, model inference, pointwise scoring and ranking, and listwise ranking. Candidate generation narrows down a vast number of possible candidates to a small set that can be ranked, while feature retrieval fetches data for the user and candidates being considered. Filtering removes candidates based on fetched data or model predictions, and model inference sends feature vectors to a model service to get predictions. Pointwise scoring and ranking scores items in isolation, while listwise ranking orders items in context with other items in the list. The goal of these components is to provide diverse and high-quality recommendations that meet user preferences and objectives.