David Kaufman

Glovo is a large delivery marketplace in Europe with 150,000 merchant partners. In addition to an app for their in-store staff, I recently led the design and development of a new portal for store managers. It provides merchants access to metrics and insights, menu management capabilities, and the ability to run promotions/ads, amongst other functionalities. The project started from zero and took a lean MVP approach in scaling to minimize the time between development, user discovery, and iteration.

I converted a standard web app into a more performant downloadable PWA for our merchant partners in store operations. The improvements brought them push notifications, the offline mode, smaller download size, native-like intractability, and improved UX.

Implemented the integration with Google Translate to translate Tweets into the users' preferred language across iOS, Android, and the web. Negotiated terms from a business standpoint and led the design and technical implementation. A/B testing showed strong improvements in the time spent, an increase in conversations (replies), and Tweet engagement.

I led the research, design, and implementation of a unique short-form video functionality on BeReal, marking a significant expansion of the platform's media formats beyond photo sharing. BeReal was primarily known for its photo-sharing capabilities, but user engagement was plateauing. The platform needed to diversify its content offerings to maintain and grow its user base. I was tasked with leading the project from conception to execution, conducting market research to understand user needs, designing the video feature, and overseeing its integration into the existing platform.

I led the research, design, and implementation of an AI-powered tool to allow accountants to extract capital transactions from PDF files automatically. The tool used OCR to read PDFs from account statements and tax documents, then LLMs to extract, clean, and label output into a structured format for manual review and further processing. The tool ended up reducing manual time spent on data extraction and document processing by 45% and greatly minimized manual human errors.

I designed and built an AI-powered tool from scratch to consolidate and derive product prioritization insights from large qualitative datasets of product usage and qualitative user feedback. The tool helped the team objectively select the best new features to build for our users based on projected impact and confidence level while minimizing engineering effort. The output provided not only a ranked roadmap of high-priority features but also supporting evidence from the user data and market analysis in order to justify the investment.

In the end, the project increased the success rate of positive A/B tests of new features from 45% to 70% while reducing the time spent in roadmap planning by roughly 30%.