Paras Doshi

This is a neat trick. It is obvious that materialized views improve performance of the query when used. What is not obvious is that you DO NOT need to reference materialized view in your query to benefit the performance improvement (query optimizer can do it for you).

9

3 Comments

Snowflake buy the dips and wait Conclusion and Recommendations The Snowflake story is not "a year out"—it's unfolding now. The company delivered 29% growth in Q3 FY26, raised full-year guidance to 28% growth, crossed $100M in AI revenue, and accelerated RPO growth to 37%. These are not the metrics of a story that's delayed. However, the full potential may still be 6-12 months from materializing. Snowflake Intelligence, Adaptive Compute, and the channel strategy are all in early innings. If you believe in the long-term thesis (data platforms powered by AI are critical infrastructure), starting a position now makes sense rather than waiting for more confirmation—by which point the stock may have already priced in much of the upside. My recommendation: Invest now with a 12-24 month time horizon. The combination of re-accelerating growth, emerging AI monetization, stable net retention, and improving margins suggests the risk/reward is favorable. The "year out" framing was relevant 6-9 months ago; today, the company is executing and the evidence is mounting that the strategy is working. For more cautious investors, a phased approach makes sense: establish a partial position now to capture the current momentum, then add on evidence of sustained AI adoption, NRR stability above 120%, and successful channel scale. But don't wait for all the evidence—by then, the opportunity may have passed. "We are executing with urgency and focus and maintaining deep partnerships with our customers that enable us to capture the opportunity in front of us and sustain durable momentum." — Sridhar Ramaswamy, SNOW Q3 2026 The data supports this statement. The question isn't whether to invest—it's whether you're comfortable with the consumption model's inherent variability and believe the AI-driven data platform thesis will play out over the next 2-3 years. If yes, the time to act is now, not later.

20

The Databricks Startups team answers founders’ most common AI questions—where to start without hype, how to stay compliant with sensitive data, what to build versus buy, how to cut hallucinations, and how to keep costs in check. 💡 Get step‑by‑step starting points, governance patterns for sensitive data, a pragmatic build‑versus‑buy lens, tactics to curb hallucinations, and cost controls that scale. 🔗 Full post: https://lnkd.in/ehRa3Yp5 #Databricks #DatabricksForStartups #AI #GenAI #MLOps

63

Are Your Data Leaders Thinking… or Just Connecting? Here’s an uncomfortable truth. Many data teams spend most of their time not analyzing… not predicting… not innovating… But connecting systems. Fixing flows. Maintaining pipelines. That’s not a talent problem. That’s an architecture problem. When ingestion, governance, and analytics live in different tools, Your smartest engineers end up stitching instead of strategizing. That ratio is broken. Modern platforms like Microsoft Fabric shift the balance: • Built-in ingestion • Built-in governance • Built-in analytics Less system stitching. More business thinking. Because scaling insight shouldn’t require scaling headcount. 📩 datafabric@quadranttechnologies.com Write to us for a quick walkthrough or POC discussion. And if you’re heading to FabCon Atlanta (March 18–20), let’s connect in person; see you there. Microsoft | Quadrant Technologies | Ram Paluri, MBA | Vamshi Reddy | Bhaskar Gangipamula | James Kass | Prakash Nagarajan | Lavina DSilva | Mithun P N | Dr. Madhavi Gundavajyala | Siva Kanuru | Sivani Pamidi | Varsha Panguluri | Ajay kumar Erukulla | #Microsoft #DataStrategy #MicrosoftFabric #CIO #CDO #ModernData #FabCon

91

2 Comments

Skills over headcount Hiring more data scientists isn’t the only lever. Enabling Citizen Data Scientists (CDS) might deliver value faster. Enter CDS - Domain experts + low/no-code = workable models and insights. - They handle “simple to moderately sophisticated” analytics and pass the best ideas to DS for production. Operating model - Educate: core analytics concepts, basic stats, reading model outputs. 📘 - Equip: AutoML, metadata management, data glossaries, self-service marketplaces. ⚙️ - Enable: sandboxes to test safely without touching production. 🧯 Outcome - Higher throughput of experiments. - Better alignment between business needs and DS work. - DS time shifts to advanced science, not “1+1=2” requests. 👉 Question: If you launched a CDS program next quarter, what’s the first capability you’d teach? Bonus - How it works: 1️⃣ Governance 2️⃣ Controls 3️⃣ Education and limits ... the goal is giving the right tools at the right moment for the right tasks. 🔗 Full article published on Medium (link in comments)

7

2 Comments

At Snowflake, we act as #CustomerZero when it comes to the use of our own products -- for one, because they're that good 😊 , and also so we can accelerate product innovation and reinforce data trust prior to public release. We want our products to be the best they can be, so we test and provide feedback to our product teams when we identify even the smallest opportunity for improvement. I spoke about this and how to overcome legacy data silos, accelerate enterprise AI adoption, and build trustworthy systems with SiliconANGLE & theCUBE at #SnowflakeSummit. Check it out: https://lnkd.in/gKWm6s_s

651

2 Comments

Quantization in Large Language Models (LLMs) As LLMs scale, intelligence is no longer the bottleneck — memory, latency, and cost are. Quantization turns out to be one of the most powerful (and misunderstood) techniques enabling real-world deployment of large models. In this blog, I break down: • Why LLMs are surprisingly quantization-friendly • INT8 vs INT4 vs NF4 — what actually works in practice • How modern architectures absorb low-bit noise • Where quantization fails and how to design around it Quantization is no longer an optimization step — it’s foundational infrastructure for production LLM systems. Link: https://lnkd.in/gqGje75G #LLMs #AIEngineering #Quantization #MachineLearning #GenAI #MLOps

9

2 Comments

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.

396

8 Comments

Mini Project: RAG Pipeline Using LangChain on Databricks By now, I’d read a lot about chunking, embedding, vectors, and RAGs - all the theory. But I wanted to get hands-on. Sure, RAG might be “old school” and there are plenty of low-code/no-code tools out there replicating its functions. But for me, this project wasn’t just about building a RAG pipeline. It was about truly internalising the concepts, applying them within Databricks, and using LangChain myself. It’s one thing to read about it, and another to get it working end-to-end. So here’s a simple tutorial to help you follow along - read PDF documents and ask questions from them using a RAG pipeline on Databricks. https://lnkd.in/dYT_AJrN #RetrievalAugmentedGeneration #RAG #LangChain #Databricks #AI #MachineLearning #DataScience #VectorSearch #DataEngineering #TechTutorial #HandsOnLearning #AIProjects #DataAnalytics #MLModels #BigData #Innovation

38

15 Comments