engblogs

Databricks-powered self-optimizing football defense analysis chatbot that learns from domain-expert feedback to continuously improve prompts, judges, and data tools via an end-to-end DC Assistant workflow built on the Databricks Agent Framework, MLflow, Unity Catalog, and GEPA-based optimization.

A technical tour of how data engineering designs and maintains scalable, quality-driven data pipelines—from ETL/ELT and storage choices to real-time analytics and AI readiness.

Delta Lake Explained shows how to turn flexible cloud data lakes into reliable, governed lakehouses with ACID transactions, time travel, and scalable performance for analytics and ML.

A concise technical overview of modern data integration, covering ingestion, transformation, loading, governance and orchestration to unify disparate sources into a trusted, scalable analytics platform.

Foundation-first strategies unify data, governance, and architecture to accelerate enterprise AI with semantic context, simplified architectures, and responsible agentic capabilities.

A concise technical overview of the Codex app, outlining its architecture, core features, and potential use cases.

A technical overview of how Snowflake and OpenAI collaborate to bring frontier intelligence to enterprise data through AI-powered data platforms.

A technical deep dive into re-architecting the ad-serving stack to support lightweight ranking models for next-gen ads.

DiffSyn, a diffusion-based generative AI, predicts diverse synthesis routes for complex materials like zeolites, enabling rapid, multi-path design and accelerating the materials discovery process.

Agentforce Enhanced Chat delivers an agent-first, highly customizable web experience with backward-compatible migration for 3,000+ tenants via a one-click upgrade/downgrade and AI-assisted tooling to accelerate development and deployment.

A cloud-native, single-tenant HSM offering hardware-isolated key management with quorum-based administration, revocation controls, and CMEK-enabled integration to meet stringent regulatory standards while simplifying deployment.

Serverless GPU-accelerated inference on Cloud Run enables on-demand deployment and autoscaling of massive LLMs using the NVIDIA RTX PRO 6000 Blackwell GPUs.