Akeel Din, PMP, CSM

Excited to share a new blog that I co-authored with Sachin Kulkarni and Sayan Pal, now published on Medium! We explored how organizations can build a governed Data Product Marketplace using automated metadata enrichment powered by IBM watsonx.data. The blog covers key architecture concepts, governance practices, automation workflows, and how watsonx.data simplifies scaling enterprise data platforms. If you’re working with data engineering, governance, or AI adoption, this article may help you gain clarity on modern data product strategies. #IBM #WatsonxData #DataGovernance #DataMarketplace #MetadataManagement #DataProducts #AI #DataEngineering

11

Very insightful analysis on why IBM made a $11B big bet on Confluent! This isn’t about data streaming, it’s about who controls the nervous system of Enterprise AI Andrew Wells Charlie Doubek Rajesh Awasthi Deepak Satya Brian R Becker Joe Cupano Jeremy Stano Sri Srinivasan Justin Becker

8

New Snowflake project: Dallas electricity demand analysis ⚡   Loaded ERCOT hourly data into Snowflake   Built SQL for seasonality, YoY trends, anomalies, and daily cycles   Pivot heatmap highlights summer peaks across years   Repo: <https://lnkd.in/gmiZJ-kJ> #Snowflake #SQL #DataEngineering #Texas #ERCOT

13

If you’re watching AI move from demos to decisions, keep an eye on multi-agent systems (MAS). Instead of one monolithic model, you orchestrate a team of specialized agents—planner, researcher, optimizer, critic—each doing a focused job and coordinating toward an outcome. Why this matters: many real-world problems are cooperative by nature. Think power grids balancing demand, ports scheduling ships, or cities timing traffic lights. MAS mirrors how these systems actually work: many actors, partial information, dynamic constraints. A simple field guide - Roles: Planner (goal + task graph), Researcher (evidence/tools), Optimizer (constraints/OR solver), Critic (verification/safety), Executor (actions/integrations) - Workflow: Sense → Plan → Negotiate → Act → Verify - Coordination patterns: Blackboard (shared memory), Market/Auction (bidding), Hierarchical (controller + workers) Where this gets practical - Climate & energy: agent “retailers” and “producers” balancing supply/demand in simulations like PowerTAC point to smarter grid strategies (http://www.powertac.org/) - Logistics: port/yard/vehicle agents plan and re-plan around disruptions; pair agents with OR-Tools for routing/scheduling (https://lnkd.in/gCEFSW6r) - Traffic management: decentralized agents adapt signals in simulators such as SUMO (https://lnkd.in/gjvKk4up) - Resilience: multi-agent allocators triage resources across hospitals or shelters under constraints How to start without boiling the ocean 1) Pick a thin-slice outcome (e.g., ETA prediction + re-routing for a single lane or route) 2) Define roles and a minimal contract: what each agent sees, says, and decides 3) Choose orchestration: LangGraph for explicit state machines (https://lnkd.in/gPPGXEt3) or AutoGen for conversational coordination (https://lnkd.in/g-3xnmxT) 4) Add guardrails: tool whitelists, budget/time limits, critics for safety checks 5) Test in simulation first: Mesa for agent-based modeling (https://lnkd.in/gyHE6crp) or PettingZoo for multi-agent RL environments (https://lnkd.in/gUfeKh3W) 6) Observe everything: logs, cost-per-decision, time-to-decision, error taxonomy What’s next - Agent-to-agent protocols will standardize tool use and negotiation - Digital twins become the default sandbox before hitting production - KPIs shift from model accuracy to decision latency and reliability under change Curious where MAS could fit your stack? What’s the smallest proof-of-value you’d try in 30 days? Share your ideas or challenges below—let’s learn from each other. Follow for more such content. #ArtificialIntelligence,#Technology,#Innovation,#Business,#Sustainability,#MultiAgentSystems,#AIforGood,#Logistics, #TrafficManagement

2

AI is creating a seismic shift in business, enabling AI-native companies to hit major revenue milestones up to three times faster than legacy firms by automating core workflows and unlocking new services-based pricing models. #AI #AgenticAI #AITransformation https://lnkd.in/ghsQ-39p

9

IBM fell about 13% today — widely reported as its worst single-day drop since 2000. This wasn’t about a quarterly miss. The market reacted to something structural. Anthropic’s new Claude Code capabilities focus on analyzing and modernizing legacy COBOL systems — automating work that historically required years of specialized effort. COBOL still underpins critical infrastructure. Estimates suggest it handles ~95% of U.S. ATM transactions and remains deeply embedded in financial systems. This isn’t just about IBM. When AI understands complex systems faster than you do, advantage shifts. - Russell Sarder, February 23, 2026 #AI #IBM #Anthropic

36

16 Comments

Coming soon to our Informatica Dallas user group, we’ll explore what it really takes to connect Agentic AI to the kind of data that actually drives decisions. The premise is simple: AI is only as valuable as the governed, high-quality data it can act on. Without a single, consistent entry point, AI agents end up chasing fragmented sources, making inconsistent calls, and delivering outputs you can’t trust. We’re building a live demonstration using Informatica’s IDMC platform to show a persona-centric Co-Pilot for a business user. This Co-Pilot is powered by a Model Context Protocol (MCP) server that lets AI agents plug directly into Informatica’s CDGC, CDQ, and MDM environments. It’s not about adding another chatbot layer—it’s about giving AI secure, policy-aware access to governed data assets, with the ability to search, discover, and trigger workflows in context. >>>More..Coming soon

29

2 Comments

Since James Malone posted something about Snowflake & turned off comments so he doesn't respond to anyone having a different take, I'll comment by reposting it. 1. Storage is cheap when you don't need a ton of it. PB is about $270K a year so I am sure many companies would love to save 30% of that cost by storing it in more performant & compressed format & which they do. Tables in this format happens to be fully managed by Snowflake(easier), have no way to access the data files directly by anyone(more secure) & have 4-5 levels of encryption that get rotated every 30 days and rekeyed in once a year(even more security) They give better performance & security & require no maintenance. Perfect for sensitive datasets or if you prefer not to maintain a lake. 2. if Customers don't want that, with Snowflake they have the option to use lakehouse via Iceberg & Parquet on their own tenant. This is called giving customer an option to pick their own storage method or even mix it up. Malone believes everyone should go opensource(which Delta is not) since it is the ONLY way his product can store & manage data. Per Malone it sounds like Whether customer loses some performance, has more maintanence, pay more for storage and own the responsibility of securing & auditing of all files & access is not important enough as interoperability. Pretty sure not everyone would agree to that. Unfortunately not everyone wants to maintain thousands of files and folders themselves & Be responsible for the file security on top of data access security via the query engines for the sake of interoperability. Sometimes, they just want to load data, query it, use ML & AI on it and not be bothered with upkeep. They want the easy button. Not everyone wants or needs Interoperability but if they do Snowflake offers it and does not force everyone to manage their own storage and that is a really good thing even if Malone thinks otherwise. Comments open for anyone. This is an open discussion.

77

10 Comments

Last night's presentation featuring David Krakov from Honeydew (Powered by Snowflake) at this month's DFW Snowflake User Group explored an interesting question: Why do organizations need independent semantic platforms separate from their BI tools and data platforms? His answer was compelling. These platforms enable organizations to define business terms, metrics, and dimensions in one central location that serves both traditional BI and emerging AI/LLM applications. As David highlighted, the modern data team's mission extends beyond shared reporting and includes providing a unified semantic model that enables diverse consumption of governed, trusted data. I'm seeing the need firsthand while implementing Snowflake's Cortex Analyst functionality. We're building semantic models distinct from existing BI and transformation tools. This requires an investment of time and also creates risk for diverging definitions. Ideally, we could enable AI on structured data in a way that leverages our existing semantic investments (especially in BI platforms). With Honeydew and other semantic platforms, we can potentially maximize our investment of time and get better governance benefits. David demonstrated the Honeydew Slack application and LLM functionality and explained how the product also works with leading BI tools as well. The vision is clear, but the execution requires careful consideration of implementation tools and approaches tailored to specific business contexts. I'd love to exchange ideas and experiences with anyone else who is looking to define and utilize best practices with semantic tools, so please let me know if you want to discuss further! #DataSemantics #Snowflake #BusinessIntelligence #AI #DataGovernance

30

4 Comments

Of the top 10 fastest-growing jobs in Dallas, look at #2: AI consultants and strategists. A role that barely existed a year ago is now one of the hottest in the market. The best way to build the skills for this kind of work? AI-Native Foundations and AI-Native Change Agent from AI-Native Training by Scaled Agile, Inc. It’s a strong signal that the work Asim Javed, Rashid Smith and I have been laser-focused on over the past few years is showing up in the market — and fast. https://lnkd.in/dY2CNC6X

29

An agile, responsible, and efficient approach to AI development and deployment, leveraging smaller, specialized models and intelligent automation through AI agents. #ai_agents https://lnkd.in/g3qBKCqF

2

Dashboards show problems; the DII AI Assistant solves them: traces the full path, correlates signals across the stack, pinpoints root cause + recommends fixes in seconds. Domain-specific AI built on decades of expertise. Private preview customers love it, they call it game changing, and now this capability is in Public preview. Look at Praveen Vijayaraghavan's post below to learn more!

24

1 Comment

⚡ Designing Enterprise Data Platforms for the AI Era I recently sat down with Louis Landry, CTO of Teradata, who's leading a significant transformation at the $1.75B revenue cloud analytics company. We explored how his team is redefining enterprise platforms to empower AI and accelerate value from data. 🔹 Louis explains why platforms must evolve and how AI requires enterprise context 🔹 Mixed workloads enable speed and scale for AI systems 🔹 From centralized to hybrid: Teradata's multi-cloud transformation 🎧 Episode highlights: — Architecting hybrid platforms to support AI agents — Insights into vector search and model-sharing — Teradata's role in large-scale enterprise ecosystems — Trends reshaping the AI + data landscape 👉 Listen here: https://lnkd.in/efkDgnPy 💬 How are you adapting your data architecture to serve AI initiatives? #AI #EnterpriseData #CloudAnalytics #Technovation #DigitalTransformation #CTO #DataPlatform

30

2 Comments