Hybrid catalog search

At WIZEWERX, I worked on a supplier catalog spanning over 100,000 product categories. Searches took more than 30 seconds, making query latency a practical problem for the application.

I built a hybrid search system with Typesense, combining keyword and vector search. Text matching and semantic retrieval worked together to find relevant catalog results.

After the change, searches returned in under a second.

I also developed full-stack analytics and reporting features in a distributed consultancy team. My work at WIZEWERX made extensive use of Ruby on Rails, Next.js, TypeScript, MobX, and MongoDB, alongside Docker Swarm and Traefik.