MasterClass
with Joe Reis
DAY 3; MAY 8 | 10:00
Mixed Model Arts:
Data Modeling in the Age of AI
Your data warehouse was designed to serve dashboards.
Now it’s supposed to serve dashboards, an AI recommendation engine, a real-time pricing service, and a semantic search platform.
Date | May 8
Location | W7 Amphitheatre, Kistamässan
Duration | Half Day
Price | 875 EUR per attendee
*exclusively available only to Data Innovation Summit delegates
If you’re being honest, you’re probably already mixing modeling approaches – a half-baked star schema here, a shoddy Postgres database there, maybe some wide tables for the ML team. Most experienced practitioners figured out a while ago that no single paradigm covers everything. The problem isn’t that people are doing data modeling wrong. The problem is that there’s no structured way to think about how different modeling approaches fit together, when to reach for which one, and how to keep the whole thing maintainable as AI workloads pile on new demands.
That’s what this Masterclass is about.
I’ve spent the last few years writing a new book on data modeling. This isn’t another rehash of dimensional modeling or normalization theory, but a practical framework for the reality most teams are already living in: multiple modeling paradigms, multiple consumers, and AI changing the requirements faster than the architecture can keep up. I call the framework Mixed Model Arts, and in this masterclass, you’ll put it to work.
What we’ll actually do during the MasterClass:
You’ll work in small teams on a realistic scenario. Here, a company where four different consumers (BI analysts, an AI recommendation engine, a real-time pricing service, and a semantic search platform) all need fundamentally different things from the same underlying product data. You’ll use the most current AI tools throughout as a practical part of the workflow, for mapping and pressure-testing requirements, generating and comparing schema designs across paradigms, translating models between styles, and stress-testing your architecture against changing business needs.
Along the way, you’ll map requirements before anyone draws a schema, model the same domain across dimensions, document, graph, and AI-native patterns, design an integrated architecture under a real constraint (a small team that can’t maintain infinite complexity), and use LLMs as a sparring partner to critique your designs and catch blind spots.
The point isn’t to hand your modeling work to a chatbot. It’s to learn where AI genuinely accelerates your thinking, and where your own judgment and experience remain what matters most. After all, data is a thinking person’s sport.
What you’ll take back to work
A requirements mapping template for assessing modeling needs before committing to a paradigm. An architecture evaluation scorecard for grading multi-paradigm designs on consumer coverage, evolution cost, AI-readiness, and operational complexity.
A 30-day cross-training plan for building multi-paradigm fluency.
All things you can apply to whatever you’re working on right now.
Who should be in the room?
Data engineers, analytics engineers, data architects, and technical leads who’ve built at least one analytical data model and are feeling the friction of serving AI workloads with patterns designed for a BI-only world.
You should know SQL. Bring a laptop with access to an LLM (ChatGPT, Claude, Gemini, whatever you prefer). No setup or installation required.
Your results WILL vary depending on the model. This is also part of the Masterclass.