Thu, Jun 18 · 6:30 PM CEST
## Details
⚠️ Please remember to include your full name and ID number (ID card, passport, etc.) in the registration form. You will be required to present this document upon arrival to access the iLAB .
Welcome to ML with Flow ! This is a new space designed for technical profiles looking to build Machine Learning and GenAI systems with a focus on reliability, transparency, and best practices.
We are kicking off this series of meetups with a clear goal: to share practical knowledge, real-world experiences, and frameworks that help bring AI into production —professionally, yet in a relaxed and collaborative environment.
Agenda:
18:15 Registration and Welcome!
Abstract
In this talk, we explore how to build an AI-in-the-middle MLOps layer with MLflow where our current architecture, large language models, and human experts work together to manage and evolve the ML lifecycle end to end. Using our production MLOps setup as a real-world case study, we will show how MLflow underpins experiment tracking and the model registry, while an AI assistant sits in the middle to orchestrate workflows, enforce guardrails, and surface actionable insights.
We will walk through three connected topics:
A deep dive into our current MLOps architecture and how MLflow integrates with cloud platforms, data pipelines, and operational tooling.
How we use AI to accelerate the migration of models from Amazon SageMaker to Databricks , including assisted code and config translation, environment parity checks, and validation of migrated models.
How to embed AI-in-the-middle across the full model lifecycle , from experiment design and deployment decisions to monitoring, incident triage, and continuous improvement loops.
Attendees will learn how to transform this idea from backlog concept to production-ready workflow , with practical patterns for using MLflow and AI inside CI/CD, migration playbooks, monitoring and governance flows, and the day-to-day operations needed for stable, compliant ML at scale.
Bio:
Carlos Rosado is Machine Learning Operations Lead at dLocal , where he focuses on building scalable, production-grade MLOps platforms to run machine learning reliably in production. With extensive hands-on experience industrializing ML workloads, he is a regular speaker at international AI and data conferences , including Databricks’ Data + AI Summit in San Francisco and the Data + AI World Tour.
