engblogs

A production-ready case study of how BGL built a secure, scalable AI agent for analytics using Claude Agent SDK on Amazon Bedrock AgentCore to democratize business intelligence through a strong data foundation and domain-aware skills.

A wearable ultrasound imaging project from SMART and MIT enables 48-hour continuous real-time cardiovascular monitoring at home through bioadhesive coupling, AI-assisted diagnostics, and cross-institution collaboration to transform chronic disease care.

Examines ROI in the public sector through agentic AI and AI agents, highlighting productivity and security gains, Gemini for Government deployments, and the essential role of executive sponsorship in scaling adoption.

Technical examination of English quirks that challenge language teaching—covering verb transformations, adjective order, and countability—to explain why rules are hard to articulate and how to improve pedagogy.

Cloudflare R2 Local Uploads enables near-client writes with asynchronous cross-region replication to the bucket, delivering faster, strongly consistent uploads across regions.

Agentic AI-enabled SageMaker Data Agent within Unified Studio accelerates healthcare data analytics by turning natural-language clinical questions into structured, executable plans and production-ready SQL/Python/PySpark code, enabling rapid cohort and survival analyses while preserving governance and security.

Practical guide to building production-ready enterprise AI agents with Amazon Bedrock AgentCore, detailing lifecycle engineering, observability, tooling, automated evaluation, security, and scalable multi-agent patterns.

Leverage Amazon Quick Suite's custom action connectors to securely upload text files to Google Drive through an OpenAPI-based integration powered by API Gateway and AWS Lambda.

Databricks Lakebase is GA on AWS as a serverless, compute–storage‑decoupled operational database that unifies analytics, real-time workloads, and AI capabilities with on-demand scalability for production apps.

MemAlign introduces a lightweight dual-memory framework that aligns LLM judges to domain-specific feedback via Semantic Memory and Episodic Memory, enabling rapid, low-cost improvements from natural-language feedback without updating model weights, and achieving memory scaling with open-source MLflow and Databricks support.

An exploration of how Databricks combines OKR-centric delivery, pod-based agile deployment, embedded expertise, and AI tooling to accelerate value and measurable outcomes for customers.

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.