Pablo Fernandez

Genuinely curious — what does your data modelling practice look like today if you're running on Databricks or Snowflake? 🤔 I've been thinking about this a lot lately. When I started out, tools like Erwin were *the* standard. Visual ER diagrams, forward and reverse engineering, a proper metadata repository. It gave data modelling a real seat at the table when most organisations were still figuring out what a data warehouse even was. But that was an era of: → Monolithic on-prem warehouses → Slow, deliberate schema design cycles → A small group of data architects owning all the models Now we're in a world of lakehouse architecture, medallion layers, streaming ingestion, and far more people touching data than ever before. Does Erwin — or the category it represents — still have a meaningful role? Or has the practice of data modelling fundamentally shifted around it? Because on modern Databricks and Snowflake stacks, the conversation seems to have fragmented: 🗂️ **Cataloguing & lineage** → Unity Catalog, Snowflake Horizon, Atlan, Secoda, DataHub, Alation 📊 **Semantic modelling** → Cube, AtScale sitting between the warehouse and BI 🧱 **Data vault & dimensional modelling** → still alive, but increasingly expressed in code rather than GUI tools ✏️ **Diagramming** → quietly shifted to dbdiagram.io, Lucidchart, or even just Notion So genuinely asking — for those building on either platform today: 👉 Do you have a formal modelling practice, or has it become emergent through your catalog and pipelines? 👉 Is anyone still using Erwin, erwin DI, or ER/Studio and finding real value? 👉 Or has the discipline quietly dissolved into the engineering workflow? No agenda — just trying to understand where the craft is actually heading. 👇 #DataModelling #Databricks #Snowflake #DataEngineering #DataArchitecture #LakeHouse #DataGovernance #ModernDataStack #DataCatalog

6

📐 What Do Data Architects Earn in the UK in 2025? With businesses scaling faster and data stacks growing more complex, Data Architects are in high demand—and their salaries reflect it. 💡 In our latest guide, you’ll find: ✔️ Average UK salary benchmarks for Data Architects ✔️ How location, industry, and cloud expertise impact earnings ✔️ What senior-level architects can expect in 2025 If you're hiring for this role—or planning your next career move—this breakdown gives you a clear view of what’s competitive. 👉 Read the full report: https://lnkd.in/e7Q7P3C7 Hiring data talent this year? Let’s talk about how Live Digital can help. #DataArchitect #TechSalaries #DataCareers #HiringInTech #LiveDigital #UKTechJobs

1

This was one of my first projects after joining GeoDS, and my first paper in the world of Geography/Geographic Data Science! We needed census data at small area geographies across the whole of the UK for another project (which will follow soonish...). Putting it together from the different national sources into a unified dataset, and validating that we were combining like-for-like variables, turned out to be pretty time consuming. So we've published it as an open dataset. If you have a project which requires census data across UK national boundaries, this will hopefully save you some time.

25

The UK public sector is stuck in a data mess. Not because of a tech shortage — but because it still runs on scripts. The future? Autonomous data agents. They don’t just automate tasks. They achieve outcomes. → Clean data → Faster insights → Less manual work Engineers stop coding. They start commanding. This is the agentic shift. #AgenticAI #AIagents #PublicSectorInnovation #GovTech #DataOps #DigitalTransformation #AutonomousAI #AIinGovernment #AIforGood #SmartData #RplusAnalytics

15

2 Comments

There is an interesting news for you, if you are working on Time Series forecasting! Google’s updated Time Series Foundation Model now brings a much longer context window (great for long-range dependencies and regime shifts) and state-of-the-art accuracy. On the GiFT-Eval benchmark, TimesFM-2.5 ranks #1 across all accuracy metrics among zero-shot foundation models — without task-specific fine-tuning. What’s new & why it matters - Stronger long-context reasoning → better forecasts for complex signals (finance, energy, climate, supply chains, healthcare). - Lean + capable: ~200M parameters, decoder-only, native probabilistic/quantile forecasting for risk-aware planning. - Easier access: weights live on Hugging Face; BigQuery / Model Garden integration is on the way. - Reliability focus: longer memory + improved inference options help stabilize results on noisy, non-stationary data. Bottom line: Foundation models for forecasting are moving from proofs-of-concept to production-ready tools. If your team plans demand, balances grids, manages inventories, or tracks climate risks, TimesFM-2.5 should be on your shortlist. #TimeSeries #Forecasting #AI #MachineLearning #GoogleResearch #TimesFM #GiFTEval #DataScience #MLOps #BigQuery

4

2 Comments

In Kantar 's recent webinar, we explored the capabilities of True Signal. Unlike traditional tracking methods that rely on rolling averages and can delay change detection, True Signal delivers today's data by removing outliers and using AI to provide the best estimate of current metrics. 📈 This approach offers more sensitive and immediate insights into marketing effectiveness, acting as an early warning system to help brands make timely decisions and course corrections. True Signal can be applied to various data types, including survey, search, social, and sales data, providing a comprehensive view of both short-term movements and long-term trends. Learn more about bridging the data-decision gap by watching the webinar on-demand: https://lnkd.in/eyp8qGWf #BrandIntelligence #DataIntelligence #Marketing #Tracking

3

We're joining with Databricks and Women in Data® to help close the gender health gap with GenAI - find out how 👇 Kubrick Associate Director Mirela Gyurova set out to see if Generative AI could help uncover bias in Lupus research - a condition of which 90% of patients are women.   What she found was a way to expedite research for underfunded areas of medical research while improving the governance of critical human decision-making.   In this webinar, Mirela takes us under the hood of her GenAI tool, exploring the challenges of LLM training, stringency, and best practice to help mitigate biases, rather than exacerbate them.   Learn how her Multi-Class Classification model leverages AI self-reflection, accuracy measurement, and human input to uncover a dramatic disparity in bias detection and mitigation in health research. Register for your place: https://bit.ly/3JGzmfb #AI #GenerativeAI #Webinar #WomeninData #WomensHealth #WomensHealthGap #AIforGood

62

2 Comments

The 'Participation Trophy' Era of AI is Over The headline statistic from McKinsey’s State of AI 2025 report is that while 88% of organisations have adopted AI, only 6% (the "Vanguard") are seeing significant financial returns. The remaining 94% are stuck in "pilot purgatory," sprinkling AI on legacy processes rather than reworking them. We have entered the era of Agentic Darwinism, defined by three shifts: - Generative to Agentic: We are moving from chatbots that write to autonomous agents that do (executing multi-step workflows with minimal oversight). - IT to Strategy: For the Vanguard, AI is a CEO-led strategic imperative, not an IT support ticket. - Efficiency to Innovation: The 94% use AI to cut costs, whilst the 6% use it to build entirely new business models. The market is bifurcating. Organisations will either become Agentic Innovators that are rewiring for scale or remain niche, human-centric boutiques. The "middle pack", or those treating AI as a bolt-on tool, will be squeezed out, bearing the costs of tech without the speed of the Vanguard. #AI #McKinsey #AgenticAI #FutureOfWork #TechStrategy

12

What is the current state of the UK data industry? As an industry we perform comparatively well against other sectors. Despite this, there is still a way to go until the opportunities of Data, Tech & AI are fully realised.  This research aims to inform policy, industry strategy and talent development by identifying gaps and guiding future investment. We’re asking:    →How far have we come?  →Where do barriers still exist?   →What are the must have skills in the sector?  →And how are these challenges shaping long-term career and economic outcomes?    This is not just about women’s experiences, It’s a way of understanding our collective voice across the industry to consider your future workforce needs.     📩TAKE THE SURVEY: https://lnkd.in/eR9sGBFh This Women in Data® Annual State of the Nation Survey 2025 is powered by Yonder Consulting. #WomenInData #SOTN

26

Head of data job market I’ve spoken to 40 ish heads of data this year alone And every single one had multiple opportunities on ie interviews So there is clearly an abundance of roles being advertised or looked for But there are 2 things that are of significant interest First thing, the conversion rate is low, ie people are in play with 4/5 companies to land one offer. I mean interviewing, presentations etc. CVs sent out and applications will be a lot higher But why? Another element is the role type, some are very commercial heavy and others data domain specific. Ie data science could be multiple domains now AI, ML and analytics/reporting. Counting out people who don’t have the specificity. Job specs and therefore CVs to match these specs aren’t hitting what’s exactly needed. But there’s enough in the CV to warrant a discussion. As a head of data, it’s never been more important to showcase the value, the impact and the commercialisation of your recent roles. Data is under a different spotlight but it still seems hiring process hasn’t shifted to match.. Looking for a new role, or need a specific head of data? The markets changing, are you? #data #recruitment #hiring

10

So many of you have been asking about this. Here is another stellar way to bring live data viz, tables, dashboards straight into Chronicle with Equals Bobby Pinero Tejas Gawande Shekhar Kirani

42

1 Comment

Is data quality overrated? James Garner thinks so. Yes, clean data matters. But modern AI — especially large language models — are astonishingly good at handling messy, inconsistent inputs. So if you’re using data quality as an excuse to delay your AI projects… you might be missing out. In this episode of Data & AI Mastery, we explored why “less-than-perfect data” shouldn’t be the thing that holds you back — and how experimenting now puts you far ahead of waiting for a mythical clean dataset. 🎧 Watch the full conversation here:

8

1 Comment

Gabriele Gnecco ~ The 4 stages of data adoption at a football club: 1. Outsourced analysis: Uses raw data but rely external consultants/ providers. 2. Internalised analysis/ external engineering. 3. All data phases are internalised, but datasets analysed in separate silos. 4. Holistic approach: The holy grail! Data is centralised, unified and analysed. Hudl Paul MacLeod Jon Ollington Alex Marin Felices 🔗 https://lnkd.in/eZriJN4N

27

3 Comments

With all the buzz about Databricks' Discover right now, thought I'd share this demo I recorded with Paul Roome (Product @ Databricks), showcasing the value of Discover for finding the right business data.

46

1 Comment

AI in analytics in 2026 is going to force an awkward truth:   Semantic modelling isn’t going anywhere.   For the last couple of years, a lot of the AI narrative has implied semantic modelling matters less.   In reality, AI just turns your existing mess into faster, more confident confusion. If you don’t have shared definitions, you don’t have “talk to your data”,  you have “talk to your ambiguity”.   2026 is where more orgs will accept that semantics is the unlock: - metric definitions that are explicit and versioned - grains and join paths that are intentional - certified sources, lineage, and permissions baked in - a single place to answer “what does this KPI actually mean?”   The twist is that the UX gets much better: - instead of forcing everyone into a BI tool, AI becomes the interface - the semantic layer becomes the contract underneath it - analysts spend less time re-explaining definitions and more time on decisions   Hot take: AI doesn’t replace semantic modelling. It makes it finally worth the effort, because you get leverage across every surface area: dashboards, self-serve, reports, and chat.   If you’re investing in analytics AI this year, I’d start here: build the governed metric layer first, then add chat on top of it.   Agree or disagree?

8

3 Comments

So this happened: Tableau Desktop for $0. Great for individual analysts or small teams who need to visually explore their data, wherever it lives, to gain quicker insights and understanding. https://lnkd.in/eGbqhxY8

96

5 Comments