Darshil Mehta

Inference shall be a differentiating factor among the Data Scientists and an Important Element to focus in the Enterprise level data science and decision making. Even for any organisation that cares about the evidence that could answer the business question need to emphasise the Inference (simple statistical inference to causal inference), Statistical Inference is something helps in building better predictive models, This is the right time that “Noise in the Data Science Community is settling down”, and giving space for such dialogues and unique opportunities! A new practice of DS needs to be “Evidence and Inference” focused! Even if we want to understand the patterns of how adapted Gen AI systems helps in the business problems , how to perform evals of an Agent or a RAG solution - Inferential Mindset is somethting critical! ps - just thought of recording my thoughts and ideas as a video for better communication and engagement! 🙂‍↔️

130

8 Comments

Rubbing shoulders and sharing knowledge with Australia's best Data Governance practitioners in Sydney last week was so insightful. 💡 It is a complex area that is evolving rapidly to keep up with the expansion in Data and AI Some key points that were made clear: - AI won’t succeed without strong data foundations. - Before you invest in models, invest in data governance. It ensures your data is accurate, consistent, and compliant. - Without it, AI delivers noise. With it, AI delivers value. Start with governance. Build trust in your data. Then scale with AI. 👉 Is your data ready for AI? Message me to find out more. #RUBIX #DataGovernance #DataEnablement #AI 🤟 PS: Huge congratulations to the amazing Natalie Hogan for winning Data Governance Leader of the Year! 🏆 👏

56

7 Comments

When I first came up with the topic for this report, I wasn’t exactly pumped. Some people love it, but for me - data governance has always been a bit of a snoozefest. But the more I researched, debated, and wrote, the more I realised this may just be the fundamental problem at the heart of GIS. For most spatial data, governance isn’t built in - it’s clunky, slow, often an afterthought. I’ve seen even the most governance-minded orgs fall back on spreadsheets to track data records. Good governance should be seamless, built-in and accessible - a thing of intrinsic value, not a lead weight. This matters more than ever, as AI transforms geospatial analysis and we can’t rely on humans to hand-hold with governance caveats at every step. That’s what my new report with CARTO & Snowflake is all about: how to make spatial data governance future-proof in the age of AI. Download it here! 👉 https://lnkd.in/eR3Qx3-x #GIS #geospatial #datagovernance #datamanagement

66

2 Comments

An important update for colleagues delivering #datagovernance here in NZ: the guidance for Māori data associated to the new Information Sharing Standard has now dropped. When the initial guidance came out the Māori data governance requirements embedded in the new standard led to a few second order questions. The initial supporting guidance developed the material in the standard, but that was only provisional. That guidance has since been chopped up into sections and expanded, E.g model legal clauses for updating contracts. I started analysing the implications of the four small rules in the standard in the beginning of June and only concluded about three weeks ago. Aka the implications are significant for the existing body of data governance regimes (broadly understood). Furthermore, the expectations are embedded in the standard so we can't pick and choose the bits we like/agree with and treat the remainder as optional Fortunately there are pathways to making all the various regimes synchronise but obtaining that solution necessitates changes in the order of operations (there is a BEDMAS for data governance). Usefully the standard elevates the Te Kāhui Raraunga framework up a level and that excellent body of work helps with the reframing and reordering.

32

2 Comments

Your organisation might have zombie datasets. I'm talking about the datasets that are feeding into dashboard applications reports, but have no maintainers or a responsible owner. Technically, they are called orphan datasets. They can easily contain stale data that propagates through models and reports. Thereby increasing operational risk. Get rid of zombie datasets by: ➤ Enforcing ownership policies at the point of dataset creation. ➤ Setting up lineage tracking to catch abandoned sources ➤ Setting alerts on freshness thresholds The real culprit will always be the lack of a strong data ownership culture in the organisation. Until your team actually cares about who owns data, zombie datasets will keep coming back.

54

7 Comments

Getting the right data for meaningful evaluation can be challenging. Especially in an environment that values quantitative data above all. At our recent webinar, this came through strongly from participants. Many shared that: * Performance indicators can't tell the full story * What they want to understand through evaluation can feel disconnected from what they're required to report on These are challenges we see in lots of the work we support — and we’ve learned there are ways to work with them. Here’s what’s helping: ✅ Starting with purpose – being clear on why we’re evaluating and tracking impact, and what we want to learn, who we must report to ✅ Building in reflection and insight, rather than simply complying with requirements ✅ Connecting performance data with a wider understanding of change, through an evaluation framework that includes wider data ✅ Offering meaningful alternatives by supplementing with qualitative data where standard indicators fall short ✅ Taking ownership of the outcomes that matter to your organisation and people – so we can stay focused even when external frameworks shift We’re always learning from the organisations we work with, and from these kinds of rich conversations. There aren't easy answers, but we can often find ways forward. Catch-up on the webinar through the link in the chat.

29

2 Comments