Browse All Machine Learning Content (288)

Paula Santamaría and Julia Schröder Langhaeuser present a production Retrieval-Augmented Generation (RAG) architecture built on Azure Database for PostgreSQL, explaining why Postgres can be a solid foundation for RAG at scale and what it takes to move from prototype to production with performance tuning and monitoring.

Abe Omorogbe explains how PostgreSQL is evolving into a backbone for production AI agent workflows, focusing on reliable and safe data retrieval. He covers MCP-based agent patterns, common failure modes when agents generate SQL, and emerging approaches like context correction and blended retrieval across relational, vector, and graph techniques.

Varun Dhawan explains why PostgreSQL is increasingly used as an “everything database” and how its extensibility lets teams consolidate workloads like OLTP, analytics, and vector search. He also connects these capabilities to Azure Database for PostgreSQL and scaling patterns such as Citus.

amitch announces a preview ServiceNow connector that can query Microsoft OneLake data in place (zero-copy), aiming to reduce ETL and keep a single governed data foundation across Microsoft Fabric analytics and downstream operational workflows.

jiang_jenny1 introduces the Fabric Spark Operations Skill (preview), an AI-assisted, read-only troubleshooting tool for Spark workloads in Microsoft Fabric. It turns common investigations—failed notebooks, pipeline failures, session triage, and performance issues—into natural-language commands that produce a severity-ranked diagnostic report with fix recommendations and links back to Fabric.

Microsoft Developer previews Livestream 3 of POSETTE 2026 with short introductions to 11 PostgreSQL talks, including WAL, logical replication, testing and coverage, consistency in clusters, performance tuning, and a session on production RAG at scale using Azure Database for PostgreSQL.

Mohsin Ejaz explains how to build safety tooling and guardrails for automated, AI-driven PostgreSQL tuning, focusing on monitoring, validation, and risk controls so performance improvements don’t come at the cost of outages or regressions.

jiang_jenny1 announces a preview update to the Spark History Server in Microsoft Fabric that speeds up loading for large Spark applications by switching to snapshot-based, incremental rendering. The post also covers expanded support for Spark Streaming configurations like event log compression and rolling logs.

dotnet explains patterns for modernizing data and migrating line-of-business applications incrementally, focusing on moving the database first and evolving the app through stable API layers and modern data access approaches.

royr explains how Microsoft Fabric surfaces service issues and incidents, including in-product banners, Teams/email notifications, and the Service Health dashboard in the Fabric admin portal, plus what admins can configure and who can view detailed health messages.

UlrichChrist announces a preview integration in Microsoft Fabric that uses a Microsoft-provided ABAP Add-On to extract SAP data at high throughput via Copy job in Fabric Data Factory, landing it in OneLake for analytics workloads and downstream reporting and AI scenarios.

Satya Nadella highlights an Azure milestone: a new performance record for a leading LLM training benchmark at extreme scale, achieved through full-stack work across silicon, systems, networking, and software in partnership with NVIDIA.

azinh17 breaks down how Azure achieved a top MLPerf Training v6.0 result for Llama 3.1 405B, training at extreme scale across 8,192 GPUs. The post focuses on the cluster and network architecture choices—NVLink scale-up domains, Azure’s MRC fabric, and topology-aware parallelism mapping—that kept step time stable as the system scaled.

diptiborkar announces new Microsoft Fabric and Azure Databricks interoperability that lets teams use Microsoft OneLake as a shared, native storage layer, including GA read access and beta support for writing Unity Catalog managed tables. The post also frames OneLake as a governed data and context foundation for analytics and AI agent workloads.

Anavi Nahar rounds up Azure Databricks announcements and sessions from Databricks Data + AI Summit 2026, focusing on tighter interoperability with Microsoft’s data stack (OneLake, ADLS) and governed access via Unity Catalog, plus new integrations like the Excel add-in, SharePoint ingestion, and OneLake catalog federation.

preshah announces new interoperability features between Azure Databricks and Microsoft Fabric: storing Unity Catalog managed tables directly in OneLake (beta) and a “Publish to Fabric” workflow (preview) that creates mirrored catalog items from Databricks so the same tables can be queried across Fabric workloads without copying data.

Alberto_Manuel explains why telecom operators need an ontology (semantic) layer to keep data meaning intact for GenAI and analytics, and outlines how Microsoft Fabric IQ (preview) uses ontology items, graph relationships, and data agents to enable cross-domain reasoning, governance, and scalable AI-driven data access.

Natalie Guevara announces the GitHub Multilingual Repositories Dataset, a CC0-licensed metadata dataset that helps researchers find public repositories with non-English text in READMEs, issues, and pull requests, and explains what’s included, how to use it for evaluation, and where the caveats are.

John Edward explains how Microsoft Fabric OneLake in Azure acts as a single, organization-wide data lake and why it matters for modern enterprise analytics architecture, including reducing data silos, supporting lakehouse patterns, and improving governance and AI readiness.

GeertVanTeylingen outlines a zero-copy pattern for making enterprise file data usable by modern AI and analytics platforms, using Azure NetApp Files as the system of record and Microsoft OneLake shortcuts to expose that data without migration or duplication.

This week in ML is a reminder that production reliability lives in the details: licensing and entitlements in Azure AI Foundry, VM and disk changes that can reshape workloads, and the day-to-day reality of cold starts, probe timeouts, and OOM kills. We also saw practical guidance for handling regional capacity limits in Azure Databricks and for standardizing failure logs across Fabric and Synapse pipelines with Azure Monitor and KQL. On the product side, Fabric added real-time dashboard improvements, governed sharing options (including OneLake shortcuts and cross-workspace role management), and more Copilot-driven authoring paths that fit into versioned, repeatable workflows.

John Savill rounds up a week of Azure platform changes and retirements, spanning compute/storage updates, database and identity improvements, monitoring changes, and several developer-facing AI items including GitHub Copilot Agent Mode in SSMS and Azure AI Foundry agent licensing and model availability.

BhaktiRath95 walks through common failure modes when running AI/ML inference workloads on Azure Container Apps, including slow model startup, probe timeouts, OOM kills, and GPU initialization problems. The post provides concrete probe settings, Python/FastAPI patterns, and Log Analytics queries to diagnose and fix issues methodically.

MichalBar introduces a preview redesign of Microsoft Fabric Real-Time Dashboards tile editing, adding Copilot-assisted visual authoring alongside a more code-friendly KQL workflow. The post walks through creating a visual from a prompt or query, iterating with history, and testing parameterized queries directly in the editor.

MichalBar introduces the Time Series Visualization (Preview) for Microsoft Fabric Real-Time Dashboards, focusing on how to explore multivariate time series data with series search and grouping, synchronized time-range navigation, and visual customization options like adaptive scaling and linear/log axis modes.

Laurent Bugnion explains why Microsoft Build 2026 labs are a popular way for developers to learn through hands-on sessions, including tracks that cover AI, Copilot, and Microsoft Fabric. He also shares where to find the on-demand “Digital lab” sessions and how long they remain available after the event.

Dan Wahlin demonstrates an “agentic journey” workflow that takes an app idea through planning, coding, infrastructure creation, and deployment to Azure, using GitHub Copilot CLI and Azure skills to handle tasks like Bicep templates, health probes, and database wiring for an app backed by Azure SQL and Microsoft Foundry.

MichalBar announces general availability of Live Refresh for Real-Time Dashboards in Microsoft Fabric, an event-driven refresh model that updates visuals when new data is ingested. The post explains how it reduces polling overhead, supports pausing/resuming during investigation, and provides configuration options like fallback refresh intervals.

analyticanna introduces Rayfin (Preview), an open-source SDK and CLI for turning content like markdown into shareable, hosted sites deployed as first-class items in Microsoft Fabric, with stable URLs, workspace-based access control, and data captured into a SQL database in Fabric for use across OneLake workloads.

Rafia Aqil explains how to diagnose and respond when Azure Databricks clusters can’t start or scale due to Azure regional VM capacity constraints, including what to send to Microsoft support, which VM families to switch to, and longer-term design choices like instance pools, serverless compute, and multi-region deployments.

SindhuBharadwaj introduces a Fabric-first migration flow that lets you mount an Azure Data Factory instance inside a Fabric workspace and migrate selected pipelines without switching portals. The post outlines the migration steps, supported connection/authentication mappings, and the validation work to do before re-enabling triggers in production.

Aaron Merrill announces a preview feature in Microsoft Fabric’s OneLake catalog that lets admins assign, edit, and remove workspace role memberships across multiple workspaces from the Secure tab, helping teams standardize access controls and reduce drift as Fabric estates scale.

Rafia_Aqil outlines a reference architecture for ingesting both streaming and batch data through Microsoft Fabric into Azure Databricks, using OneLake/ADLS and a medallion (Bronze/Silver/Gold) layout. The post breaks down five Fabric-to-Databricks integration paths and calls out security, governance, and monitoring considerations.

Sharlkaur introduces a preview workflow for AI-authored Power BI reports in Microsoft Fabric, using Skills for Fabric and the Power BI authoring plugin optimized for GitHub Copilot CLI. The post shows how agents can generate PBIR-based reports from prompts, iterate using screenshots, and publish to Fabric as part of an end-to-end agentic analytics flow.

shiv_narayanan announces general availability of SharePoint and OneDrive Shortcuts in Microsoft Fabric OneLake, explaining how Fabric can reference Microsoft 365 files in place, optionally transform supported formats into Delta tables, and use Entra ID-based identities for production authentication and scale.

Sally Dabbah explains how to turn Synapse/ADF/Microsoft Fabric pipeline failures into structured, queryable telemetry by sending standardized failure events into Azure Monitor Log Analytics via the Logs Ingestion API and a Data Collection Rule, enabling KQL-based analysis, alerting, and reliability reporting across environments and datasets.

This week in ML, Microsoft Fabric moved closer to an agent-ready analytics platform, with new ways to ship backends into Fabric, ground agents in governed context, and model relationships directly on OneLake. Rayfin positions Fabric as a default deployment target for data-powered apps, while Fabric IQ (now GA) and its ontology support aim to standardize how agents request context with permissions and auditability built in. Graph in Fabric (GA) adds GQL-based relationship querying, and the Fabric Operations agent plus Fabric Skills show how Microsoft wants teams to monitor, automate, and code against Fabric with guardrails instead of one-off scripts.

John Savill runs through a Build-special weekly Azure update, covering a wide set of platform announcements across compute, containers, integration, monitoring, databases, Fabric/Databricks, and Azure AI Foundry—plus security-focused items like confidential computing and Purview agent integrations.

Charles Feddersen and Abe Omorogbe explain how AI apps and agents change database design, focusing on reasoning over operational data instead of only transactions. They demo new capabilities across Azure SQL Database, Azure Cosmos DB, and Azure HorizonDB (cloud-native PostgreSQL) to simplify architectures and reduce latency.

Milos Colic shares how Xoople scaled Python-based AI workloads on Azure using Ray via Anyscale, covering the distributed-systems challenges behind data ingestion, training, and inference, and why the team prioritized delivering outcomes over operating clusters.