🗒️ **About the event** Join **Grafana & Friends Manchester** at **bet365** for an evening of talks, live demos, and community conversation! We’ll explore how to build dashboards that work for your whole team, find the performance limits of an LLM service, and bring observability to a classic arcade game. Expect practical examples, ideas to try in your own projects, and the chance to put your driving skills to the test! 🏎️ 📅 **Thursday 22 October 2026** 🕕 **17:45–20:00** 📍 **bet365, Zenith Building**, just off Market Street, between Manchester Piccadilly and Victoria stations. The host team will welcome you downstairs, check you in and escort you up to the event. 📋 **Agenda** **17:45 · Welcome, food, and refreshments** 🍕 Meet fellow community members and grab a bite before the talks begin. **18:15 · Talks and live demos** 🎙️ 🎙️ **One dashboard, a thousand contexts** **Speaker: Mihaela Maior, Director Engineering, Grafana Labs** How do you build one dashboard that serves your entire team, instead of maintaining ten conflicting copies? Mihaela will explore filters, multi-property variables, and variables defined at the instance, folder, dashboard, and section levels. Through demos and practical examples, you’ll learn how to use these features to adapt a single dashboard to different contexts. 🎙️ **How much traffic can your GPU really handle before an LLM service starts to fall apart?** **Speakers: Ben Kaimogul, Senior Infrastructure Engineer at Callcare Ltd, and Ruth Kegicha, Innovation Discovery Manager** What happens when an LLM application reaches its limits? In this live, hands-on session, Ben and Ruth will put a retrieval-augmented generation (RAG) application running on vLLM and NVIDIA T4 GPUs under increasing load—and watch what happens with Prometheus, Grafana, and k6. Using tests that behave like real users, they’ll connect user experience with inference metrics to explore latency, throughput, queue depth, token generation, and GPU saturation. You’ll learn how

Description supplied by the event organiser.