About
I lead Microsoft's Azure Data Analytics, building the Fabric engines for data…
Articles by Bogdan
-
Analyst Council in Action: From "Launch AI Minesweeper" to a Full Strategy
Analyst Council in Action: From "Launch AI Minesweeper" to a Full Strategy
This is the 3rd article on Using GitHub Copilot CLI to Build a Multi-Agent Market Analysis Workflow. The first article…
2 Comments -
From One Prompt to a Team of Analysts - Building AnalystCouncil in GitHub Copilot CLI
From One Prompt to a Team of Analysts - Building AnalystCouncil in GitHub Copilot CLI
This is the second of a 3-part series on Using GitHub Copilot CLI to Build a Multi-Agent Market Analysis Workflow. The…
7 Comments -
The Tool That Surprised Me — GitHub Copilot CLI Beyond Code
The Tool That Surprised Me — GitHub Copilot CLI Beyond Code
Part 1 of 3: Using GitHub Copilot CLI to Build a Multi-Agent Market Analysis Workflow My day job doesn't give me much…
11 Comments -
Pivoting from Synapse to Fabric: Rebuilding Without Losing the Team
Pivoting from Synapse to Fabric: Rebuilding Without Losing the Team
In 2021, I moved from leading Power BI Engineering to what is now Fabric Analytics. Shortly after, I had to 'kill' the…
7 Comments
Activity
8K followers
-
Bogdan Crivat shared this#MicrosoftFabric #Spark Native Execution Engine is emerging as an amazingly performant and often surprisingly cheap solution for data engineering at any scale.Bogdan Crivat shared thisIf you’re running #Spark on #MicrosoftFabric, you’re sitting on a massive performance advantage thanks to the Native Execution Engine. This represents a fundamental architectural shift that replaces traditional JVM-based execution with a vectorized C++ path powered by #ApacheGluten and #Velox. The result is a system that handles data engineering workloads with incredible speed. By processing Parquet and Delta datasets more efficiently, you get significantly more value out of your Fabric Capacity Units. Because this is built directly into the infrastructure, you enjoy faster query times and higher throughput for your ETL pipelines without rewriting a single line of Spark code. To verify if the engine is active, check the query plan in the Spark UI. Look for node names ending in the suffix Transformer, NativeFileScan, or VeloxColumnarToRowExec. These suffixes confirm that the Native Execution Engine successfully processed the operation.
-
Bogdan Crivat shared thisImmensely proud of the innovation coming from #Azure , Microsoft Gray Systems Lab , and #FabricAnalytics . In the noisy environment of all sellers wanting to introduce new categories and convince everybody that “B is the new A”, fundamental improvements in core analytics technologies are what really will make AI fly over your data.Bogdan Crivat shared this📣 We are delighted to announce the SIGMOD 2026 Industry Track Best Paper Award Winner and Honorable Mention! Congratulations to all the authors below! 👏 👏 💐 ==================================================== CoddSpeed: Hardware Accelerated Query Processing in Microsoft Fabric Matteo Interlandi (Microsoft)*; Nicolas Bruno (Microsoft); Brandon Haynes (Microsoft); Carlo Curino (Microsoft); Rathijit Sen (Microsoft); Yinan Li (Microsoft); Kaushik Rajan (Microsoft); Bailu Ding (Microsoft); Lukas Maas (Microsoft); Wei Cui (Microsoft); Kevin Gaffney (Microsoft); Mingsheng Hong (Microsoft); Brian Kroth (Microsoft); Sampath Rajenda (Microsoft); Peng Cheng (Microsoft); Surajit Chaudhuri (Microsoft); Johannes Gehrke (Microsoft); Raghu Ramakrishnan (Microsoft); Lidong Zhou (Microsoft); Momin Al-Ghosien (Microsoft); Craig Peeper (Microsoft); Marius Dumitru (Microsoft); Conor Cunningham (Microsoft); Kevin Bocksrocker (Microsoft); Prasanna Sundar (Microsoft); Ron Boskovic (Microsoft); YongChul Kwon (Microsoft); Marko Zivanovic (Microsoft); Runbin Shi (Microsoft); Krystian Sakowski (Microsoft); Denis Lamtsov (Microsoft); Josep Aguilar Saborit (Microsoft); Krish Srinivasan (Microsoft); Adarsh Kapil (Microsoft); Andrew Putnam (Microsoft); Anudeep Kambapu (Microsoft); Aparna Konduri (Microsoft); Aaron Landy (Microsoft); Aaron Moore (Microsoft); Aaron Pitman (Microsoft); Artem Oks (Microsoft); Ashit Gosalia (Microsoft); Babu Kandimalla (Microsoft); Blake Pelton (Microsoft); Bogdan Crivat (Microsoft); Cesar Galindo-Legaria (Microsoft); Chidamber Kulkarni (Microsoft); Esteban Calvo (Microsoft); Evgeny Babin (Microsoft); Hans Lehnert Marino (Microsoft); Igor Konev (Microsoft); Israel Arroyo (Microsoft); Jesus Camacho Rodriguez (Microsoft); Luis Penaranda (Microsoft); Mandar Datar (Microsoft); Morten Borup Petersen (Microsoft); Nicholas Simmons (Microsoft); Parminder Poonian (Microsoft); Pravija Danda (Microsoft); Rob Rydberg (Microsoft); Ronak Bajaj (Microsoft); Sai Sumanth Kammal Shetty (Microsoft); Sharanya Bhat (Microsoft); Zach Iracheta (Microsoft) ==================================================== Industry Track Honorable Mention Enzyme: Incremental View Maintenance for Data Engineering Ritwik Yadav (Databricks, Inc.); Supun Abeysinghe (Databricks, Inc.); Min Yang (Databricks, Inc.); Jeffrey Helt (Databricks, Inc.); Manuel Ung (Databricks, Inc.); Yuhong Chen (Databricks, Inc.); Melody Hu (Databricks, Inc.); William Wei (Databricks, Inc.); Yiming Yang (Databricks, Inc.); Tom van Bussel (Databricks, Inc.); Sourav Chatterji (Databricks, Inc.); Indrajit Roy (Databricks, Inc.); Paul Lappas (Google, Inc.); Yannis Papakonstantinou (Google, Inc.); Tahir Fayyaz (Databricks, Inc.); Bilal Aslam (Databricks, Inc.); Ross Bunker (Databricks, Inc.); Michael Armbrust (Databricks, Inc.); Shrikanth Shankar (Databricks, Inc.) #SIGMOD2026 #ACM #sigmodindustrytrack #sigmodaward
-
Bogdan Crivat shared thisA great article regarding the #MicrosoftFabric integration with Epic https://lnkd.in/gMGwY_du "Epic Fabric delivers real value, and health systems are prioritizing it for three primary reasons: Faster access to Epic data, with refresh cycles moving from daily batches toward near-real-time insight Relief from brittle, high‑maintenance extract, transform, load pipelines that consume teams and delay insight Access to new Epic data models, with Epic planning to release all new data models exclusively through the Cogito Cloud " As the article calls out - Fabric is the catalyst for *your* analytics and AI strategyEpic’s shift to Fabric: The analytics pivotal moment health systems can’t ignoreEpic’s shift to Fabric: The analytics pivotal moment health systems can’t ignore
-
Bogdan Crivat reposted thisI ran the same query twice. Same 76 million rows. Same filters. Same Fabric warehouse. The only difference - one table had `CLUSTER BY`. The other didn't. Result: → Without clustering: 210 MB scanned, 3,771 ms → With clustering: 75 MB scanned, 1,897 ms 64% less data. 50% faster. What makes this result even more interesting - the regular table had warm memory cache. Its data was already sitting in RAM before the query ran. The clustered table was fetching fresh from remote storage. Even with that advantage, the regular table was slower. That's the power of File Skipping. When clustering is set up correctly, the engine knows exactly which files to open - and skips everything else. Less data touched = faster results, every time. This matters beyond just query speed. Slow storage means slow dashboards. Your BI team re-runs the same reports. Users lose trust in the numbers. All because the engine is scanning data it doesn't need to. One line of SQL at table creation fixes this. I wrote a full step-by-step guide covering: → What Data Clustering actually is (storage level) → Which columns to cluster on and which to avoid → How to implement it with real SQL → How to validate results using Query Insights All with real numbers from my own test run - not benchmarks. Microsoft Fabric,Microsoft, Peri Rocha Link in the comments 👇
-
Bogdan Crivat reposted thisBogdan Crivat reposted thisProud of how KPMG is continuing to modernize how we serve clients. By re-platforming KPMG Digital Gateway’s data layer on Microsoft Fabric, we’ve further unified data, analytics, and governance in a single place. The result: data layer onboarding time decreased 87%, allowing our teams to meet client needs even faster. Learn more about the effort: https://lnkd.in/g5QibWE5
-
Bogdan Crivat reposted thisBogdan Crivat reposted thisWill Power BI ever properly support third party semantic models? The answer is no for a variety of technical reasons, so if Power BI is your front end of choice then Power BI semantic models are where you should store your metrics. https://lnkd.in/eGufMDiQ
-
Bogdan Crivat shared thisMore #Microsoft magic around #Github #Copilot #CLI! With #Dataverse's skills, and #Fabric Skills (https://lnkd.in/gQbY7_fH) you can now fully access core data assets running on #Azure in your favorite agentic toolBogdan Crivat shared thisYour coding agent just learned Dataverse! Dataverse Skills is a new open-source plugin that gives GitHub Copilot and Claude Code native, end-to-end knowledge of Dataverse: connect, build, and operate, from a single prompt. Describe what you want. The agent handles environment setup, schema design, data loading, and querying. No tool flags. No context switching. This is what it looks like when enterprise platforms become agent-operable. 👉 Try it today https://lnkd.in/gFXTVkWU #Dataverse #AIAgents #PowerPlatform
-
Bogdan Crivat reposted thisBogdan Crivat reposted this🚀 Join us for the Online Microsoft Fabric Roadshow – Livestream | Tuesday, May 19, 2026 You might just be able to catch me in Action in the keynote with Chris Webb!! Can’t join the Microsoft Fabric Roadshow in person? Tune in to the livestream and don’t miss out. During this one‑day online broadcast, business decision makers and data professionals come together to explore the latest innovations in data, analytics, and AI with Microsoft Fabric — bringing the energy and insights of FabCon and Microsoft Ignite straight to your screen. You’ll gain practical insights, real‑world use cases, and the latest Fabric updates, all live and interactive. Whether you’re shaping data strategy or building solutions, this livestream helps you unlock more value from data and AI — wherever you are on your Fabric journey. 👉 Register now and join the livestream from anywhere: https://msft.it/6048QIygr
-
Bogdan Crivat shared thisIn my last post (https://lnkd.in/gqWGXTP2) I have explained how I used #GitHub #Copilot #CLI to create a "council" of analysts to help prepare product strategies. In this last article of the series, I am reviewing the result of their work, and trying to figure out what worked very well and what needs improvement.Analyst Council in Action: From "Launch AI Minesweeper" to a Full StrategyAnalyst Council in Action: From "Launch AI Minesweeper" to a Full Strategy
-
Bogdan Crivat liked thisBogdan Crivat liked thisSpent the week thinking about something that keeps coming up in client conversations. Companies want AI. They want it fast. They want RAG agents pulling from internal data, Copilots embedded in workflows, custom assistants for every team. And then someone in the room asks the quiet question. Who actually has access to all this? Most of the time, the answer is uncomfortable. OneLake Security going GA in Fabric changes that conversation. Instead of stitching permissions together at every layer, you define them once on the lake. RBAC, RLS, CLS, all enforced through Entra ID. Then everything above inherits. Every engine, every report, every agent. A Foundry agent pulling from OneLake will only ever see what the user behind it is allowed to see. No custom plumbing. No security as an afterthought. This is the kind of foundation work that does not make headlines but quietly makes AI projects deliverable in regulated industries. Wrote up my take on why this matters. #MicrosoftFabric #OneLake #DataGovernance #EnterpriseAI #OneLakeSecurity #SecurityWhy is OneLake Security the best thing that happened to AI?Why is OneLake Security the best thing that happened to AI?
-
Bogdan Crivat liked thisFun time at Evolver office and meet the cool people.Bogdan Crivat liked this𝐄𝐯𝐨𝐥𝐯𝐞𝐫 𝐢𝐬 𝟏𝟖 𝐦𝐨𝐧𝐭𝐡𝐬 𝐨𝐥𝐝. We are now 85 people across the U.S., Canada, Europe, and India, bringing together 𝐝𝐞𝐞𝐩 𝐝𝐨𝐦𝐚𝐢𝐧 𝐞𝐱𝐩𝐞𝐫𝐭𝐬 𝐚𝐜𝐫𝐨𝐬𝐬 𝐭𝐡𝐞 𝐂𝐅𝐎 𝐚𝐠𝐞𝐧𝐝𝐚 with 𝐞𝐱𝐜𝐞𝐩𝐭𝐢𝐨𝐧𝐚𝐥 𝐞𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐬 and 𝐀𝐈 𝐫𝐞𝐬𝐞𝐚𝐫𝐜𝐡𝐞𝐫𝐬. 𝐈 ❤️ 𝐭𝐡𝐢𝐬 𝐭𝐞𝐚𝐦. Domain experts and technologists sitting side by side, building 𝐩𝐥𝐚𝐭𝐟𝐨𝐫𝐦 𝐜𝐚𝐩𝐚𝐛𝐢𝐥𝐢𝐭𝐢𝐞𝐬 𝐭𝐨 𝐭𝐚𝐤𝐞 𝐨𝐯𝐞𝐫 𝐜𝐨𝐫𝐞 𝐞𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞 𝐟𝐮𝐧𝐜𝐭𝐢𝐨𝐧𝐬 and deliver 𝐦𝐞𝐚𝐬𝐮𝐫𝐚𝐛𝐥𝐞 𝐯𝐚𝐥𝐮𝐞 𝐟𝐨𝐫 𝐅𝐨𝐫𝐭𝐮𝐧𝐞 𝟏𝟎𝟎𝟎 𝐜𝐮𝐬𝐭𝐨𝐦𝐞𝐫𝐬. I re-energize every day at Evolver (even after 3 days of 5 am meetings), because the 𝐦𝐢𝐬𝐬𝐢𝐨𝐧 𝐢𝐬 𝐜𝐥𝐞𝐚𝐫 (make customers successful) and the people genuinely enjoy 𝐰𝐨𝐫𝐤𝐢𝐧𝐠 𝐭𝐨𝐠𝐞𝐭𝐡𝐞𝐫. We keep growing: https://lnkd.in/gXbAQ7fX
-
Bogdan Crivat liked thisBogdan Crivat liked thisAI can do virtually anything, but the people who get the most out of it aren’t the deepest specialists — they’re the ones with enough range to know when it’s doing a good job. Here’s a simple test. If you ask AI to say something in French and you don’t speak French, it’ll probably get it right. But you don’t actually know. You just hope. That’s fine for French. It’s a problem when it’s your security architecture. Or your legal language. Or your financial model. AI is a multiplier. But a multiplier applied to blind trust doesn’t give you leverage — it gives you faster mistakes at greater scale. I spent the last two months building a personal ops pipeline — data architecture, coding standards, PII redaction, executive reporting. The reason it worked isn’t that I’m deep in any one of those areas. It’s that I know enough across all of them to push back when something’s off. That’s the thing the 10,000 hour rule never accounted for. Extreme depth in one discipline made sense when you were competing against other humans. Now you’re guiding a system that already has the depth. What it doesn’t have is your judgment across the full picture. David Epstein wrote about this in Range — the idea that in complex unpredictable environments, generalists outperform specialists. He was talking about humans competing against humans. That was 2019. The environment just got a lot more complex and a lot more unpredictable. The argument didn’t get weaker. It got more urgent. And before you say it — no, specializing in AI itself isn’t the answer either. AI plus two other disciplines beats pure AI every time. The value isn’t in knowing the tool. It’s in knowing what to do with it across domains that actually matter. Extreme specialization used to be an advantage. Right now it’s starting to feel like a liability. #AI #Range #FutureOfWork
-
Bogdan Crivat liked thisBogdan Crivat liked thisFor the first time ever, I’m telling the whole story of what happened to my career almost 10 years ago. In this week’s episode of The Knowing Moments. Nine years ago, I was offered the chance to lead one of the most important businesses at Microsoft - Identity. It was my dream job, everything I had been waiting for. But it was a role that I felt completely unqualified for. From the outside, it looked like a promotion. From the inside, I went to sleep most nights wondering if I'd be found out. The strange thing is — that was one of the most important periods of my career. Not because I figured it out quickly. But because I learned what real growth feels like. As I am embarking on my new chapter, this story reminded me I’ve been here before! Just like the moment I jumped into 27 degree ice water in Antarctica! You don’t always feel ready, you jump in anyway. Full episode 5 Out of My Depth is out today, link in comment -> #TheKnowingMoments #CourageToChange #KnowYourMoment #WhatComesNext #Leadership
-
Bogdan Crivat reacted on thisBogdan Crivat reacted on thisNew in Copilot Cowork: mobile, skills, and plugins. Now available on iOS and Android, so you can delegate work from your phone, pick it back up on your desktop, and keep tasks moving without breaking flow. And with new connectors, Cowork can operate across business systems and data.
-
Bogdan Crivat liked thisBogdan Crivat liked thisI really admire papers written like this. Because behind all the symbols, definitions, and theorems, mathematics or physics is still a deeply human activity mostly trying to understand structure, meaning, and some times maybe even our place in the universe. A playful opening like this does not weaken the rigor of the paper. It reminds us why rigor matters, because even the most abstract objects are part of our very old habit of asking impossible questions “carefully”. Also, beginning with “Since the dawn of time…” and ending up at Hodge–Riemann cohomology classes is simply beautiful. Honestly, maths and mathematical physics remain one of the most elegant ways humans have found to be serious and funny at the same time. https://lnkd.in/dTKYG7Fp
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
-
-
-
-
-
-
-
-
Licenses & Certifications
Publications
-
Fuzzy ARTMAP rule extraction in computational chemistry
IEEE - . International Joint Conference on Neural Networks, 2009. IJCNN 2009
Patents
-
Random access in run-length encoded structures
Issued US 7,952,499
Random access to run-length encoded data values is provided. A target value is identified by a logical index into a structure of run-length-encoded values. To access the value, a bookmark is selected based on the logical index, on a maximum logical index of the bookmark, and on a specified bookmark distance. An initial run in the structure is located, based on the selected bookmark. A final run is chosen, at most one bookmark distance from the initial run. The target value is the value of the…
Random access to run-length encoded data values is provided. A target value is identified by a logical index into a structure of run-length-encoded values. To access the value, a bookmark is selected based on the logical index, on a maximum logical index of the bookmark, and on a specified bookmark distance. An initial run in the structure is located, based on the selected bookmark. A final run is chosen, at most one bookmark distance from the initial run. The target value is the value of the final run. Efficiency heuristics are used when generating bookmarks or creating the structure of run-length-encoded values.
Other inventors -
-
Techniques for Evaluating Recommendation Systems
Issued US 20090319330
Other inventors -
-
Efficient column based data encoding for large-scale data storage
Issued US 8,108,361
The subject disclosure relates to column based data encoding where raw data to be compressed is organized by columns, and then, as first and second layers of reduction of the data size, dictionary encoding and/or value encoding are applied to the data as organized by columns, to create integer sequences that correspond to the columns. Next, a hybrid greedy run length encoding and bit packing compression algorithm further compacts the data according to an analysis of bit savings. Synergy of the…
The subject disclosure relates to column based data encoding where raw data to be compressed is organized by columns, and then, as first and second layers of reduction of the data size, dictionary encoding and/or value encoding are applied to the data as organized by columns, to create integer sequences that correspond to the columns. Next, a hybrid greedy run length encoding and bit packing compression algorithm further compacts the data according to an analysis of bit savings. Synergy of the hybrid data reduction techniques in concert with the column-based organization, coupled with gains in scanning and querying efficiency owing to the representation of the compact data, results in substantially improved data compression at a fraction of the cost of conventional systems.
Other inventors -
Languages
-
Romanian
Native or bilingual proficiency
-
French
Limited working proficiency
-
English
Native or bilingual proficiency
Recommendations received
5 people have recommended Bogdan
Join now to viewOther similar profiles
Explore more posts
-
Infostrux Solutions
What happens when you pair LangGraph’s flexibility with Snowflake’s secure, scalable compute? You get a seamless path to production-grade LLM applications. Fabian Hernandez, part of our Infostrux team, walks you through how LangGraph agents can now be executed entirely inside Snowflake using its new LLM function framework. It’s an exciting step forward for anyone building: -LLM agents -Retrieval-augmented generation pipelines -Conversational workflows directly tied to enterprise data This article is a must-read for teams operationalizing AI where it matters most—in the cloud, with governance built in. Explore it here: 👉 https://hubs.ly/Q03ykwlj0 #Snowflake #LLMFunctions #LangGraph #AIWorkflows #GenAI #EnterpriseAI #Infostrux
-
Palantir Technologies
The AI Forward Deployed Engineer is here. Amplify your development work with AI labor, and reach your next breakthrough faster than ever. At DevCon 3, watch as Palantir Software Engineer Ankit Shankar unveils the AI Forward Deployed Engineer — an agentic system that can build data transformations, create and edit your ontology, create functions, build entire applications, and much more.
37 Comments -
Agentize
DevCon 3: Palantir’s AI Revolution in Action Just experienced Palantir’s DevCon 3 (July 2025) - and the future of enterprise AI is here. Two days of breakthrough demonstrations that will reshape how we think about automation. The Game-Changer: AI Forward Deployed Engineer Palantir unveiled their AI Forward Deployed Engineer - an agentic system that autonomously builds data transformations, creates ontologies, and constructs entire applications. As Software Engineer Ankit Shankar demonstrated: “any clicks that humans do in the UI could be things that you can have an AI FDE do for you”. This isn’t just automation - it’s true AI labor that amplifies development work. The Results Speak Volumes Lennar is building 50x faster on AIP. Fifty times. This isn’t about efficiency - it’s about fundamentally transforming software development and business processes. Beyond the Hype: Real Enterprise Value While Palantir accepts Bitcoin and developed “Foundry for Crypto”, DevCon 3 showcased something more significant: practical AI delivering immediate business value. The AIP Deployment at Scale presentations demonstrated how organizations can confidently scale AI without disrupting production. The Agentic Future Chief Architect Akshay Krishnaswamy discussed “enterprise autonomy” - digital architectures shared between human and AI teams. Companies like Enhans demonstrated LAM-based commerce solutions, showing AI systems that actually perform tasks and execute decisions. This perfectly aligns with platforms like Agentize.eu pushing European AI automation boundaries. The convergence of enterprise-grade platforms with innovative European solutions creates powerful opportunities for organizations ready to embrace autonomous business systems. The Bottom Line We’re beyond AI experimentation into practical, scalable implementation. With AI FDE, 50x development speed improvements, and growing agentic automation ecosystems, we’re witnessing truly autonomous business systems. For business leaders: the question isn’t whether AI will transform operations, but how quickly you can adapt to leverage these breakthrough capabilities. The race to AI-powered business transformation has begun. Are you ready? #AI #AgenticAutomation #PalantirDevCon3 #EnterpriseAI #FutureOfWork #BusinessTransformation
-
StackAI
We’re proud to partner with Groq to power enterprise AI agents with low-latency inference for performance-critical workloads. Teams on StackAI deploy Groq for latency-sensitive workflows: high-volume document analysis, customer-facing interactions, and internal knowledge retrieval. In each of these situations, fast, reliable responses materially improve user experience and productivity. StackAI and Groq are a prime example of how matching the right inference provider to the right workload helps enterprises build AI agents that perform in real production environments. Want to learn more? Read the full piece at the link in comments. #StackAI #Groq #EnterpriseAI #AIagents
4 Comments -
DAIR.AI
New research from Databricks. It's about training enterprise search agents via RL. KARL introduces a multi-task RL approach where agents are trained across heterogeneous search behaviors, constraint-driven entity search, cross-document synthesis, and tabular reasoning. It generalizes substantially better than those optimized for any single benchmark. KARL is Pareto-optimal on both cost-quality and latency-quality trade-offs compared to Claude 4.6 and GPT 5.2. With sufficient test-time compute, it surpasses the strongest closed models while being more cost efficient.
8 Comments -
BOT Consulting
Every data story has a journey, but too often, we lose track of it in the noise of pipelines, dashboards, and ETL logs. In this post, Sanyam Shah breaks down how to map end-to-end Data Lineage in Snowflake, turning metadata into visibility, compliance, and trust. If your goal is to make your data traceable, auditable, and AI-ready, this is a must-read. 👉 https://lnkd.in/gspE7Qs3 #BOTConsulting #DataLineage #Snowflake #DataGovernance #DataEngineering
-
Archetype Consulting
This foundational checklist from Augusto Rosa’s Snowflake Architecture Series is a must-read. Whether you're new to Snowflake or scaling your cloud data platform, this checklist breaks down the essential building blocks—like identity and access, cost governance, and data modeling—that every data team needs to get right. Set yourself up for long-term success with Snowflake—this is the perfect place to begin. https://lnkd.in/eYP8jZ2s
1 Comment -
Snowflake
Engineering at Snowflake means solving problems no one has tackled before. For Principal Software Engineer Hyungtae Kim and the Cortex team, that meant designing GPU-powered infrastructure flexible enough to handle any AI workload, at scale. Get a behind-the-scenes look at how the foundation was built, what makes it different, and how it’s powering the next era of GenAI: https://bit.ly/4kKjGVI
2 Comments -
StorageReview.com
Data silos are one of the biggest roadblocks to efficient analytics and AI workflows. MinIO’s new Aistor Table Sharing is designed to tear down those barriers by enabling direct, on-premises data access from Databricks without copying or duplicating data. Key takeaways: • Aistor Table Sharing lets Databricks query on-prem MinIO data directly, eliminating data movement. • This approach reduces complexity, storage sprawl, and latency for hybrid and on-prem use cases. • By keeping data in place while supporting modern analytics and AI consumption, organizations can accelerate time to insight. Full story: https://lnkd.in/e4mh22md #DataManagement #HybridCloud #AIAnalytics MinIO
-
Operant AI
Data is in the air! With back-to-back Snowflake and Databricks conferences in SF, one thing that resonated is that data is paramount for AI. While everyone's talking about data use in AI innovation, securing data in the age of AI couldn't be more critical. This week, our CTO, Priyanka Tembey, dives deep into why runtime protection is essential for securing data in use within modern AI systems. Discover how Operant's AI Gatekeeper simplifies this complex challenge, making robust AI security achievable: https://lnkd.in/gYcbixSB #AIGatekeeper #SecureAI #AgenticAI #SnowflakeSummit #DataAISummit #DatabricksSummit #AIPrivacy #DataInMotion #Runtime #DataCloud #AIReady #DataTransformation #RealTimeAI #Datasecurity
2 Comments -
AtScale
Legacy semantic models were never built for the velocity and complexity of AI-driven analytics. They assume fixed schemas, predictable joins, and a handful of BI queries, not hundreds of dynamic, multi-domain requests from copilots and agents. AtScale’s AI-powered modeling engine changes that through an adaptive feedback loop. It continuously analyzes warehouse metadata, query logs, and lineage to infer semantic relationships and hierarchies automatically. When candidate metrics or joins are discovered, they’re surfaced for human validation, accepted, rejected, or refined, and those outcomes feed back into the model’s learning system. All of this runs on AtScale’s composable semantic architecture, governed through Semantic Modeling Language (SML). Every suggestion, update, and approval is version-controlled, reversible, and fully auditable. It’s modeling that scales with your data, not against it. 🔗 Learn more:
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content