engblogs

NVIDIA Nemotron 3 Nano 30B MoE model with 3B active parameters and a hybrid Transformer-Mamba architecture is now generally available in Amazon SageMaker JumpStart, enabling open-weights and fully managed deployment to accelerate AWS-based AI development.

AI is transforming data analytics by embedding AI/ML, NLP, and automated workflows across the data lifecycle to accelerate data preparation, insight generation, and real-time decision-making while balancing governance and ethical considerations.

A practical guide to understanding and choosing between structured, semi-structured and unstructured data, outlining storage architectures from data lakes to lakehouses and data warehouses, with decision frameworks and real-world use cases for actionable analytics.

Unveils the Brickbuilder Partner Network to unify and accelerate Databricks' ecosystem in the agentic AI era, aligning incentives, specialization, and demand generation to turn data and AI solutions into rapid business impact.

A concise, technically oriented narrative of Anna Sulkina's journey to Airbnb, highlighting career milestones and lessons learned in tech.

A practical guide to using Codex to design and orchestrate autonomous agents in an agent-first technology landscape.

Just-in-Time Tests (JiTTests) generated by large language models enable rapid, regression-focused testing in agentic development, eliminating test maintenance and manual code reviews.

A practical blueprint for using Gemini and GenAI to generate scalable, production-grade code samples, detailing the architecture, tooling, and seven actionable takeaways for reliable, end-to-end sample generation.

A concise synthesis of Dropbox's executive roundtable on AI-driven engineering productivity, emphasizing leadership alignment, measurable impact, tooling adoption, and the human factors shaping product velocity.

Explore two practical paths for model adaptation - fully managed fine-tuning on Vertex AI and customizable LoRA-based tuning on GKE - to produce production-ready, domain-specific AI using Gemini, Llama, and related open-source models.

Explains how LinqAlpha deploys a multi-agent Devil’s Advocate on Amazon Bedrock (Claude Sonnet 4.0) to rigorously pressure-test investment theses with structured, source-linked counterarguments.

Swann scales Generative AI to millions of IoT devices with Amazon Bedrock, detailing intelligent notification filtering, a tiered model strategy (Nova Lite/Pro, Claude Haiku/Sonnet), edge and serverless processing, and a cost- and latency-aware architecture for context-aware security alerts.