Software Engineer
CurrentCurtsy is an app that makes it easy for women to buy and sell clothing from their phones.Key Contributions:• Developed an automated clothing moderation system using machine learning techniques, including tabular classification, image classification, and natural language processing using LLMs. This system efficiently identifies items requiring manual review by humans, resulting in a 70% auto-approval rate and instant listings on the app. This improvement eliminated the previous 4-6 hour approval process for each item, enhancing operational efficiency and accelerating the item listing process.• Led the complete rebuild of Curtsy's legacy web application by transitioning from vanilla JavaScript and jQuery to React and Next.js, aligning it with our iOS app's capabilities. Utilizing Next.js features like Automatic Static Optimization, code splitting, and prefetching, we achieved a 2.5x boost in page load speeds. Accessibility improved by 45%, and the app now adheres to best practices. This upgrade increased development speeds fivefold, enabling the launch of capabilities previously only available on iOS, such as web-based selling and personalized item feeds.• Architected Curtsy’s Closet Import feature, enabling seamless wardrobe imports from other clothing resale marketplaces. This system uses a queue architecture and custom-trained LLMs for generating detailed item descriptions to help boost item sell-through rates. This strategy enabled the import of 6.4 million items in the first year, ensuring wardrobes are current across all marketplaces and improving user experience.• Served as product lead for Curtsy's Resale-as-a-Service initiative, a platform enabling clothing brands and retailers to launch their own resale marketplaces. This initiative encourages sustainable practices, reflecting the expanding focus on eco-consciousness within the fashion industry.