Join us for this in-person MLOps Community London session with amazing talks 😎

🕒 6:00 PM - Welcome chats with pizza and drinks 🍕🍺
Kick off the evening meeting fellow MLOps practitioners, founders, and engineers.

🕒 6:30 PM - Real‑Time Model Predictions with Databricks Online Feature Stores
A concise walkthrough of how to deliver low‑latency, production‑ready feature retrieval using Databricks Online Feature Stores, focusing on architecture, data design, and the operational patterns needed for real‑time model serving.
🎤 Tyler Reichardt (Senior Machine Learning Operations Engineer @ First Central)

Tyler Reichardt is a Senior Machine Learning Operations Engineer at First Central, with experience across data science, engineering and MLOps. Originally from South Africa and now based in London, he works across the full data lifecycle, from building pipelines and machine learning models to scaling reliable production systems. He is particularly interested in applying data to sport, finance and environmental challenges, and enjoys building side projects such as trading and betting algorithms. With a background in rugby, he brings a collaborative approach to solving complex problems.

🕒 6:50 PM - Async Agents in Production: Failure Modes and Fixes
As models improve, we’re building long-running, asynchronous agents - deep research agents, browser agents, and multi-step workflows that operate autonomously. But these systems fail differently from short-lived request-response agents: early mistakes compound, retries spiral, and token usage grows unpredictably. Patterns that work for simple LLM calls often break down in production. This talk covers the most common failure modes in async agents and practical design patterns to reduce error compounding and keep token costs bounded at scale.
🎤 Meryem Arik (Co-Founder & CEO @ Doubleword)

Meryem Arik is the co-founder and CEO of Doubleword, where she works on large-scale LLM inference and production AI systems. She studied theoretical physics and philosophy at the University of Oxford. Meryem is a frequent conference speaker, including a TEDx speaker and a four-time highly rated speaker at QCon conferences. She was named to the Forbes 30 Under 30 list for her work in AI infrastructure.

🕒 7:10 PM - break 🍕🍺
Refuel, connect, and continue the conversation.

🕒 7:20 PM - From Untracked Models to Trusted AI: Laying the Foundation for Scalable ML and LLMs
ML systems faced common challenges: models deployed without clear versioning, limited traceability of training data, “works on my machine” scenarios, and low trust in performance evaluations. At Mondra, we addressed these gaps by structuring MLOps workflows, introducing end-to-end traceability, and standardizing experimentation using MLflow, transforming fragmented processes into a reliable and reproducible system. This talk highlights what worked, what didn’t, and the key trade-offs behind improving model governance, boosting operational efficiency, and reducing iteration costs. The foundation also enabled the successful introduction of LLM-based product classification alongside existing models like BERT, unlocking new capabilities while maintaining control in production.
🎤 Leonardo Pedroso dos Santos (Principal Data Engineer @ Mondra)

Leonardo Pedroso is a Principal Data Engineer at Mondra, specialising in high-performance, scalable big data pipelines on Databricks. He has played a key role in building a data intelligence platform for lifecycle analysis and net zero, applying strong engineering practices to ensure reliability, accuracy, and trust in data. His work focuses on performance optimisation, platform refactoring, and production-grade deployment of data systems, alongside establishing AI foundations using MLOps and MLflow to enable reproducible, governed, and trustworthy machine learning workflows and more recently, he has been expanding the platform with AI agents, enabling more intelligent and impactful outcomes for clients.

🕒 7:40 PM - 3 Ways in Which Product Design Is Changing, and One Way in Which It Isn't
Peter explores how the design process has undergone fundamental changes over the last five years and how, in today’s generative AI world, designing system behaviour and rules often impacts the user experience far more than the shiny UI.
🎤 Peter Koraca (Founder @ Creative Crow)

Peter Koraca is half product designer, half engineer. For the past 15 years, he has helped organisations define, design, and build AI- and data-heavy products using a prototyping-first approach. His work spans data annotation tools at Meta, a family AI assistant at PanasonicWell, Agent Studio and Computer (the enterprise agent) at DevRev, and more. He combines deep AI curiosity with product craft and occasionally engages in ultra-running.

🕒 8:00 PM - Q&A speaker panel
Join all speakers for an open discussion on MLOps, platform engineering, agents, and real-world ML systems. Bring your questions!

🕒 8:30 PM - Social 🍺 Fitzroy Tavern
(16 Charlotte St., London W1T 2LY)
Continue the conversation in a relaxed setting.

Hosted by Databricks

Databricks is a leading data and AI company and the creator of the lakehouse architecture. Its Data Intelligence Platform enables organisations to unify data, analytics, and AI on a single foundation. Databricks works with thousands of organisations worldwide to help them build, deploy, and scale data and AI applications.

MLOps Community London @ Databricks March