Querying Clinical Data: AQL vs. FHIR Search vs. SQL on FHIR

Five ordinary clinical questions put to three query languages: FHIR search, openEHR AQL and SQL on FHIR. The comparison shows what each was built for, where approximate cohort queries go wrong, and how view definitions turn FHIR resources into tables. Includes a tested SQL on FHIR pipeline in Python and C# clients for FHIR search and AQL.

.NET15 min read

Amazon S3 Vectors vs. OpenSearch Serverless vs. pgvector for RAG: Cost Floors, Query Bills and Tenant Isolation

All three AWS vector stores can now cost close to nothing at rest, but they bill and isolate tenants in very different ways. This comparison models Amazon S3 Vectors, OpenSearch Serverless (classic and next generation) and Aurora PostgreSQL with pgvector across three workload shapes, checks the S3 Vectors model against the worked examples on the AWS pricing page, and shows where the tenant boundary belongs in each.

Declarative Stream Processing: EventBridge Pipes vs. Custom Lambda Glue

EventBridge Pipes is usually sold as cheaper and lighter on concurrency than a Lambda stream consumer. Checked against the AWS documentation, both claims mostly disappear once the consumer filters on its event source mapping. This article compares the two for a DynamoDB Streams to EventBridge integration, with tested code, a reproducible cost model and a decision framework.

Zero-ETL from Aurora to Amazon Redshift: What You Set Up, What It Replaces and What It Does Not

For decades, bridging transactional PostgreSQL logs against heavily optimized columnar Data Warehouses necessitated configuring deeply expensive, brittle Apache Spark pipelines effectively mapping nightly batches. Discover fundamentally how AWS Aurora Zero-ETL flawlessly synchronizes precise transactional topologies entirely distinctly effectively inside Amazon Redshift practically under 15 seconds.

Inferentia2, Trainium or NVIDIA GPUs for LLM Inference: Work Out the Cost per Token

Throughout 2024, deploying localized foundation models onto Nvidia H100 instances was considered mandatory—creating profound financial hemorrhage for enterprise inference endpoints. By 2026, transitioning production endpoints to AWS Inferentia silicon via the Neuron compiler reliably slashes operational cluster costs by over 70%. Here is the engineering matrix for abandoning CUDA and migrating to custom AWS AI hardware.

AWS9 min read

Amazon SageMaker Unified Studio: Is This the End of Fragmented MLOps on AWS?

For years, AWS Data Scientists suffered from cognitive fragmentation, bouncing between EMR, Glue, and SageMaker Studio Classic to orchestrate a single model. SageMaker Unified Studio solves the interface crisis by collapsing massive data-lake operations, Generative AI visual chains, and JupyterLab IDEs into a solitary workspace. Here is the practitioner evaluation on migrating legacy MLOps architecture into the unified future.

Amazon Bedrock Data Automation: Intelligent Document Processing Without the Pipeline Plumbing

Intelligent Document Processing (IDP) used to mean brittle pipelines of OCR, specific regular expressions, and custom Lambdas. Bedrock Data Automation changes the paradigm: declare a schema via a Blueprint, and let generative AI extract structured data from messy, unstructured multi-modal inputs. Here is the practitioner evaluation of migrating from Textract pipelines to BDA.

Amazon S3 Tables & Apache Iceberg: Native Lakehouse in S3 — 14 Months Later

Amazon S3 Tables promised to eliminate the operational scaffolding of Apache Iceberg on S3 — automatic compaction, native REST catalog, managed snapshot lifecycle. Fourteen months after the re:Invent 2024 announcement, this is the enterprise practitioner assessment: what teams migrated from traditional S3 + Glue Catalog setups, the Lake Formation column masking gap that affects regulated workloads, and the migration decision framework for data engineering teams evaluating the switch.