Data Community

PLEASE NOTE: This webinar is hosted on Microsoft Teams. Registering on this meetup page doesn't mean you register for the webinar. Please use the link provided to sign up on Teams.

This meetup is bought to you by Data Community sponsors, Ignition Data / IRiS.

If you're building a Lakehouse and betting on AI, the decisions you make in your Silver layer will determine whether your AI initiatives actually work. This session goes straight to the technical depth: how your Lakehouse architecture maps to what AI systems need to reason reliably, where implementations fall short, and what a genuinely AI-ready Lakehouse looks like in practice.
What we'll cover:

• Why AI failures are almost always a data foundation problem, and how a well-structured Lakehouse solves them
• How DV2 structures map to AI requirements: entity identity, temporal history, relationship graphs, and versioned business definitions
• Where implementations go wrong: an honest look at the most common failure modes
• The full AI stack architecture and how DV2 fits within Medallion on Fabric, Snowflake, and Databricks
• What Phase 1 looks like in practice: one subject area, 4-8 weeks, AI capability from the first sprint

🚨 PLEASE NOTE: This webinar is hosted on Microsoft Teams. Registering on this meetup page doesn't mean you register for the webinar. Please use the link provided to sign up on Teams.

How to build trusted customer profitability insight in weeks, while the business is changing underneath you.

Every transformation program assumes the data will support the change. It rarely does. New managers arrive with a mandate to fix margin, pricing, or post-acquisition performance, and quickly discover that reports contradict each other, finance doesn't reconcile to operations, and nobody can say with confidence which customers are actually profitable.

The data has become the first transformation problem.

This session is for data engineers, analytics leads, architects, and transformation managers who have to deliver trusted insight while the business is still moving. Using a customer P&L turnaround as the running example, Neil will walk through how a medallion architecture, deliberate automation, and reconciliation discipline combine to produce decision-grade data in weeks rather than quarters. He will also discuss how to handle the awkward realities of shifting priorities, acquired entities, and accounting treatments that don't agree with each other.

You'll leave with a practical delivery model for data work under commercial pressure: how to structure the platform so each layer earns its keep, how to win trust through reconciliation rather than promises, and how to give a busy sponsor enough confidence to act without pretending the data is finished.

The goal isn't perfect BI in 90 days. It's stopping the business flying blind.

🚨 PLEASE NOTE: This webinar is hosted on Microsoft Teams. Registering on this meetup page doesn't mean you register for the webinar. Please use the link provided to sign up on Teams.

The AI industry is in a peculiar state. Benchmarks are being broken weekly. Investment is at historic highs. And yet the nagging question persists: what problem, exactly, are we solving?
Somewhere along the way, the metrics became the point. Model performance scores, token throughput, parameter counts - these stopped being instruments of progress and started being the goal. The result is an industry that is extraordinarily good at measuring itself and increasingly vague about why.
This talk argues that AI is, at its core, a data problem; and not in a technical sense. The real problem is that we've imported data's oldest failure mode into AI: optimising for what's easy to count rather than what actually matters. The perfect storm of business pressure, technological novelty, and data complexity has made it harder, not easier, to stay anchored to the original question.
The invitation is simple: go back to the problem.

--

Presenter bio:
Mike Le Galloudec is a Principal Engineer Advocate at Oakland, dividing his time between engineering agentic AI solutions and translating them for non-technologists. He recently steered Network Rail’s first production grade AI application, leveraging the Microsoft Azure stack and custom agent frameworks to turn raw data into real time decisions. A former data scientist with a Physics degree from the University of Oxford – and a neural network tattoo to keep him honest, he still enjoys fine tuning predictive models at ridiculous hours.

Beyond delivery, MLG hosts Oakland’s YouTube channel and a fast-growing tech TikTok, distilling dense data topics into 90 second aha moments for thousands of viewers. Whether advising rail, transport or utilities giants, he balances builder pragmatism with tech clarity, championing agentic AI that makes enterprise datasets think for themselves.