Abed Ajraou

Ever added more data to your model only to see performance drop? Mohannad Elhamod explores why this happens, showing how new samples can introduce bias or noise.

12

5 Comments

✨ Reflections from Snowflake World Tour London ✨ I had the chance to attend the Snowflake World Tour in London, and it was an inspiring look at where the future of data and AI is heading. The message was clear: we’re entering a new era where data platforms are evolving beyond storage and analytics into engines of intelligence. Snowflake is positioning itself at the centre of that shift — not just as a data warehouse, but as a foundation for AI, governance, and innovation. Some key themes that really resonated with me: 🚀 From potential to readiness: AI opportunities are enormous, but many organisations still face challenges with siloed, low-quality data and legacy architectures. Bridging this gap is becoming a strategic priority. 🔗 Trusted data as the foundation: Solutions like Semarchy’s native MDM integration with Snowflake show how unifying and governing core data is essential for building reliable analytics and enabling AI at scale. 🧠 AI, but contextual: Capabilities like Cortex AI and Semantic Views highlight a future where AI isn’t just powerful, but also deeply aware of the business context it operates in — unlocking more relevant insights and intelligent decision-making. ⚙️ Acceleration and agility: Innovations like Gen 2 Warehouses and Snowpark Connect for Apache Spark show how quickly the ecosystem is evolving to meet the demands of modern data products and real-time intelligence. More than anything, the event reinforced how critical it is for organisations to rethink their data strategies holistically — focusing not only on technology, but also on governance, trust, and usability to truly unlock AI’s potential. I’m walking away with fresh ideas on how we can apply these principles to deliver more impactful, intelligent, and trusted data products in our own work. #SnowflakeWorldTour #DataStrategy #AI #DataLeadership #Innovation #MDM #CortexAI #ProductManagement

8

If analytics systems fail through quiet interpretive drift, then accuracy alone can’t be the safeguard. What’s required is a different kind of system — one that makes its assumptions explicit, inspectable, and enforceable. In practice, that means: - forecasts with declared units, horizons, and decision intent - tolerances that are enforced, not implied - costs and risks modeled asymmetrically, not averaged away - validation that asks “is this safe to act on?” — not just “is this accurate?” In other words, the system has to understand what a forecast is for, not just how close it was to the truth. Until analytics systems can express and govern their own decision boundaries, trust will always be informal, manual, and fragile — no matter how good the models get.

4

What the difference between Data Governance and Data Stewardship? Data Governance sets the rules of the game. It's the strategic framework that defines the policies, standards, and roles that dictate how data should be managed, secured, and used across the enterprise. Data Stewardship is the team sport on the field. Data Stewards are the hands-on custodians who make data useful to your enterprise. They ensure data quality, manage metadata, and act as the bridge between business needs and technical standards. The takeaway? With all the AI hype, its easy to forget about having strong data foundations. Governance without Stewardship is just theory. Stewardship without Governance is chaos. You need both working in harmony to transform data from a raw material into your most valuable and trusted business asset. #DataGovernance #DataStewardship #DataManagement #DataStrategy #DataQuality #Alation

12

2 Comments

94% of UAE data leaders struggle to trace AI decisions: Dataiku report Nearly all data leaders in the UAE (94%) admit they lack full visibility into how AI systems make decisions, according to the “Global AI Confessions Report: Data Leaders Edition,” by Dataiku, The Universal AI Platform. The study, conducted by The Harris Poll for Dataiku, surveyed more than 800 senior data executives across eight countries, revealing striking governance risks and overconfidence in AI across the Emirates. The findings come as the UAE accelerates its ambitions to become a global leader in artificial intelligence, guided by its National Strategy for Artificial Intelligence 2031. Yet, the results highlight a growing tension between innovation and oversight. Only 17% of UAE respondents said they always require AI systems to “show their work,” underscoring a lack of explainability in enterprise AI. While nearly two-thirds (62%) are not confident their organisation’s AI systems could pass a basic audit of their decisions, only half have delayed or blocked the deployment of an AI agent due to explainability concerns. Despite these misgivings, 72% of UAE data leaders would still trust an AI agent to make autonomous decisions in critical business workflows, even if it could not explain its results, suggesting that rapid adoption may be outpacing responsible governance. The report indicates that this risky behaviour may stem from mounting pressure on UAE data leaders and a misalignment with executive expectations. Almost six in ten (59%) respondents believe their C-suite overestimates the accuracy of AI systems, while 64% say leadership underestimates the time and complexity required to make AI production-ready. Worryingly, nearly a third (32%) have been asked to approve an AI initiative that made them feel uncomfortable, and three in four (75%) say their company’s AI strategy is driven more by technological ambition than by business outcomes. It is likely that this behaviour is reinforced by the relatively low perceived consequences of AI failure. While globally, the majority (56%) of data leaders expect a CEO to be ousted by 2026 due to AI-related complications, the UAE has the lowest percentage (35%) of data leaders who are of this opinion. Moreover, over half (53%) say they do not feel their own role is at risk if their company fails to achieve measurable business gains from AI in the next one to two years, the highest level of perceived job security worldwide. #SidBhatia #AIinUAE #DataLeadership #ResponsibleAI #AIGovernance #ArtificialIntelligenceStrategy

11

3 Comments

Vodafone advances digital transformation and trustworthy GenAI with an ontology-driven semantic model EXFO Context is an ontology-driven semantic model that provides the foundation for Generative AI integration. It enables a natural language interface to configuration and service data traditionally locked in siloed inventory and Configuration Management DataBase (CMDB) systems. Now live at Vodafone Germany, EXFO Context models the multi-layered, multi-technology network, supporting numerous departments addressing service assurance and service fulfilment use cases. Vodafone can now evaluate the impact of faults or planned maintenance in real time through automated updates, change audits, and service impact analysis, ensuring robust change management and operational consistency. This same semantic foundation supports the evolution toward autonomous network operations. By creating a dynamic, service-aware digital twin, EXFO Context enables closed-loop automation and allows Large Language Models (LLMs) to interact meaningfully with structured network data, accelerating root cause identification, impact reasoning, and AI-driven decision-making within the Network Operations Centre (NOC) and beyond. -- EXFO is a global leader in telecom testing, monitoring, and analytics, with a proven legacy in enabling data-driven operational intelligence. The EXFO semantic digital twin platform, Nova Context, has been operationally deployed in live telecom networks for over 17 years, originally developed by Ontology Systems, pioneers in applying ontologies & knowledge graphs to complex telco networks. Nova Context creates a semantic #digitaltwin of the network and services, a unified, graph-based model that captures layered topology, customer context, and operational state. This provides the foundational layer for AI-driven automation, intelligent assurance, and data quality governance. By fusing live network telemetry with ontological reasoning, Nova Context enables: * Impact and root cause analysis through graph inference * Semantic integration across OSS/BSS domains * AI-ready, structured data interfaces for GenAI & LLM applications As operators seek to scale automation and adopt GenAI, Nova Context offers the semantic core that makes it possible. Link in comments. #Automation #DarkNOC #TelcoTransformation #VodafoneGermany #NetworkAssurance #ZeroTouch -- Meet EXFO at Connected Data London 2025! 20-21 November, Leonardo Royal Hotel Tower Bridge Join us for all things #KnowledgeGraph #Graph #analytics #datascience #AI #graphDB #SemTech #Ontology 🎟️ Ticket sales are open. Benefit from early bird prices with discounts up to 30%. 2025.connected-data.london 📋 Call for submissions is open. Check topics of interest, submission process and evaluation criteria https://lnkd.in/dhbAeYtq 📺 Sponsorship opportunities are available. Maximize your exposure with early onboarding. Contact us at info@connected-data.london for more.

11

1 Comment

🚀 RAG vs Agentic RAG vs MCP: The Next Evolution in Retrieval-Augmented Generation Retrieval-Augmented Generation (RAG) is evolving—but not all RAG is created equal. 🔹 Traditional RAG A simple pipeline: retrieve documents, augment with an LLM, and deliver to the user. Fast and straightforward—but static. Little flexibility in selecting or reasoning over data. 🔹 Agentic RAG Adds intelligence: LLM agents can decide which retrieval tools to use, iteratively refine results, and adapt dynamically. Smarter and more flexible—but still constrained by tooling and interfaces. 🔹 MCP (Modular Capability Provider) A paradigm shift: standardized context protocols and multi-provider retrieval architecture let LLMs plug into a full ecosystem. This isn’t just better retrieval—it’s a new way to interact with data systems. 💡 Why it matters: As data complexity grows, static retrieval pipelines won’t suffice. Agentic and MCP-based approaches are shaping smarter, more flexible, and enterprise-ready AI systems that can handle real-world complexity. Which approach is your team exploring to unlock the full potential of RAG? #RAG #AgenticAI #MCP #RetrievalAugmentedGeneration #AI #EnterpriseAI #MachineLearning #Innovation

904

50 Comments

💡 Data Lineage — More Than Just a Map In a world where our master, reference, and transactional data are scattered across multiple platforms and geographies, one critical question stands out: How do you really track data from its source to the moment it is acquired, stored, transformed and consumed — and trust it every step of the way? 🤔 Data lineage is not just a technical diagram. It is the full story of your data lifecycle — where it originated, every transformation it has undergone, the systems it has passed through, and the rules applied along the way. Why does this matter? Because without it: You cannot be certain your insights are accurate You risk non-compliance with global regulations You slow down change impact analysis and decision-making With it, you enable trust, compliance, and agility. It is a strategic enabler that connects the technical world with the business outcomes we all care about. At @Investec, we take pride in ensuring data lineage is a priority — because we are Deeply Invested in Our Data and know it is the backbone of trusted, data-driven decision-making. In a global, multi-platform environment, getting this right isn’t optional — it is a competitive advantage. So… how is your organisation mapping your data’s story? #DataLineage #DataTrust #DataGovernance #DataStrategy #DataQuality #BusinessIntelligence #Compliance #DeeplyInvestedInOurData #Investec

14

2 Comments

Join us in a conversation with our Senior Director - Analytics and Data Science, as he deep dives into the revolutionary Snowflake Data Cloud to understand how it complements Data Mesh - a data architectural approach. Explore how Data Mesh reduces the time and effort it takes with the traditional approach of storing data at a central approach. Watch the full podcast here - https://hubs.li/Q03JXgV90 . . . . . #DataAnalytics #BigData #DataScience #AIInnovation #DataWarehouse

14

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

𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻 𝗙𝗮𝗹𝘀𝗲 𝗣𝗼𝘀𝗶𝘁𝗶𝘃𝗲: 𝗪𝗵𝗲𝗻 𝗖𝗹𝗲𝗮𝗻𝗶𝗻𝗴 𝗠𝗮𝗸𝗲𝘀 𝗧𝗵𝗶𝗻𝗴𝘀 𝗪𝗼𝗿𝘀𝗲 #ReadySteadyData | #DataTrenches | #DataForensics Not every crime in data quality comes from bad inputs. Some come from overzealous cleaning. I’ve seen address fields “fixed” by truncating anything over 50 characters. Dates “normalised” into the wrong century. Nulls filled with zero - breaking downstream averages. This is 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻 𝗙𝗮𝗹𝘀𝗲 𝗣𝗼𝘀𝗶𝘁𝗶𝘃𝗲 — when your own remediation rules destroy valid data. Forensic safeguard (Python code below): The fix? Keep before/after snapshots. Review bulk changes. Don’t let your scripts become the criminals. 💡 The best detectives solve crimes without creating new ones. #ReadySteadyData #DataTrenches #DataForensics #DataScience #MachineLearning #AI #Automation #Python #DataQuality Gary Andrew Alastair Richard Chris Dale Christine Snyman Anthony Hubert Tom Gareth Abisgold Naz Richard Justin Laura Michael

8

1 Comment

Data modeling isn’t set in stone. Too often we see teams go all-in on a rigid star schema “because the book said so,” or flatten everything into a single giant table “for simplicity.” Both extremes have costs: rigid schemas limit flexibility, and giant tables create bloat and inconsistency. A hybrid or practical star schema offers balance. Fact tables are enriched with essential attributes (like order counts or flags) to speed up BI queries, while avoiding senseless duplication. The result: a presentation layer that’s lean, fast, and flexible. Don’t follow a method blindly. design for performance, maintainability, and business value. #DataEngineering #Databricks #PowerBI #DataModeling

119

24 Comments

Thrilled to share a comprehensive resource for data professionals! 📈 I've compiled a detailed table of data management roles based on the DAMA-DMBOK framework. This guide covers key roles, their responsibilities, tools, metrics, inputs, deliverables, suppliers, and consumers—perfect for anyone looking to deepen their understanding of data governance, architecture, security, and beyond. The table includes the P, C, D, O framework, where: 🔹 P (Plan): Roles involved in planning strategies or processes. 🔹C (Control): Roles overseeing or ensuring compliance and quality. 🔹D (Design): Roles focused on designing systems, models, or solutions. 🔹O (Operate): Roles managing day-to-day operations and technical execution. Whether you're a Chief Data Officer, Data Steward, or exploring a career in data, this table is a must-have reference to navigate the data ecosystem. Download the PDF below to dive in! 🚀 #DataManagement #DMBOK #DataGovernance #DataStrategy #CareerInData

31

"What Is Retrieval-Augmented Forecasting (RAF)? On a very high-level view, instead only leaning on what a model learned in training, RAF lets the model actively look up concrete past situations similar to the current one and use their outcomes to guide its prediction." https://lnkd.in/gH4uQbJJ

3

Why 90% of data projects fail & how to succeed Many businesses invest in analytics but don’t see ROI. Why?  Lack of clear goals → Start with “What decisions will this data support?”  Poor data quality → Clean, unified data = reliable insights.  No adoption strategy → Train teams to act on insights. Want to unlock true value from your data? Start with an outcome-first approach. #DataAnalytics #DeepInsights #DataStrategy #Technozis

4

1 Comment