Tesco have kindly offered to host us again - doors will open at **6.30pm** and we will finish by 9pm, with food and drinks available. This will be an in-person event (but we will try to record the talks and share the videos later). **Our Meetup is part of [London Search Week](www.londonsearchweek.com) \- check out the other events this week\!** Our first talk will come from Natália Sumida, a Software Engineer on the Search team at [iFood](https://www.ifood.com.br/), Latin America's largest food delivery platform: *"Beyond Retrieval: Building Flexible Search Pipelines:* *Search pipelines power critical user experiences, and their reliability depends on how easily we can model, test, and evolve each stage. In this talk, I'll share practical patterns for designing flexible search pipelines, drawn from lessons learned building search at a large-scale food delivery marketplace. We'll look at how to structure the pipeline into clear, testable components (query understanding, candidate generation, ranking, and post-processing) and how to validate each layer independently."* Our second talk comes from Simon Hearne of [Zilliz](https://zilliz.com/) on *"Lessons from deploying vector search in production : Getting a vector search demo working takes an afternoon, getting it to production takes... longer than we would like! Keeping it running, accurate and fast can then become a full-time job.* *Simon has worked on hundreds of vector search deployments on Elasticsearch, Milvus and Zilliz. In this talk he covers where things typically go wrong: measuring the wrong outcomes, deploying features without tuning for them (typically filtering and faceting), unexpected costs and read/write bottlenecks. He will use a live ecommerce dataset to demonstrate each pitfall, with all data and demos open sourced for exploration."* We'll also make time for Q&A on all things search and AI and some general networking. Our Search Meetup is organised by [The Search Juggler](www.thesearchjuggler.com),

Description supplied by the event organiser.