Matthew Denton

Clinical Decision Support (CDS) is an internal platform service offered to lab-results-reporting teams for the purpose of enriching lab reports with enhanced clinical content. As a platform engineer I worked closely with the genomics reporting teams to produce clinical content running at cloud scale, and enhanced the platform to support additional lab testing and report output.

• Took ownership of development and maintenance for cloud genomics platform.

• Enhanced, extended, and maintained "accelerator" code generation engine, (using strategy pattern) an engine which generates CQL code from clinically-written excel documentation for transpilation into an executable expression tree.

• Converted slow-scaling REST-based services to AWS Lambda implementations, with both direct lambda invocation, and SQS event-based lambda triggers.

• Implemented tree shaking optimizations for cloud CQL execution.

Lead on marketplace API and UI component of JuniperMarket.com. This site is a B2B e-commerce platform, and Juniper's flagship product. It was a few weeks from entering alpha when I joined the team in July and officially launched into open beta late November.

• Lead (including mentoring, code reviews, and performance reviews) for a geographically distributed and culturally diverse team responsible for building and launching the MVP

• Quickly got spun up on and took ownership for product searches on marketplace

• Delivered proof of concept APIs using patterns to standardize both team-owned APIs and upstream APIs the team consumed

Became subject-matter-expert for, and took complete ownership of, a responsive-transaction scheduler in the core of the platform. Designed and implemented a conversion of this scheduler from monolith with a custom-sharding solution, into redis-based microservice.

• Designed and delivered configuration management APIs and UI, cache querying APIs and tools/UI, and configuration-driven check-in/checkout module for shared resources.

• Added features and bug fixes to large transaction-scheduling system.

• Inverted legacy domain model to flip the relationship between a tightly-coupled pair of first and second-class domain objects.

• Unit tested untestable code via extracting interfaces, wrapping dependencies, utilizing containers, and factories among other abstractions.