Helping our clients build software that is maintainable and well-suited to their current and future needs, with a flair for collaboration.

• Created a service synchronizing data across two SaaS platforms’ proprietary APIs for stealth-mode compliance startup,

opening opportunity for $50M in sales. AWS Lambda, AWS CDK, Node.js, TypeScript, Zod, Hono, OpenAPI.

• Prototyped React Router fullstack app deployment to AWS Lambda as target platform. React, Remix/React-Router, Type-

Script, Node.js, AWS Lambda, AWS CDK.

• Developed a web UI for stealth-mode ERP SaaS. React, Remix/React Router, TypeScript, Node.js, shadcn/ui, Tailwind CSS.

Automated and improved the quality of matching our care professionals to our clients in need of at‐home care.

• Converted new-hire matchmaking system to an event-driven architecture. Cut the time new hires waited for work from 1 day to under 2 minutes. Implemented in Amazon SNS and SQS, with a Python worker running in EKS (Kubernetes).

• Created a self-service prototype to expose automated matchmaking outcomes to operations staff. Decreased pages by 90% and saved 10+ hours per week for on-call engineers. Prototyped using tracing in Datadog and MySQL for data storage.

• Rewrote the automated matching service's API to surface prescient matchmaking outcomes in a new UI for our operations team. Reduced the time to assign staff from 4 hours to under 30 minutes. Defined the RPC API in Apache Thrift, implemented using Flask and Python, and integrated with a React user interface.

• Integrated a machine-learning model for market-wide matching into the staffing pipeline, cutting down operations teams' time spent on long-term staffing by up to 50\% in specific markets. Created RPC API to provide market information and receive suggestions, defined with Thrift, and implemented with Flask and Python.

Improved the collection and presentation of near‐real time data from manufacturing plants to process engineers.

• Refactored a critical-path data streaming pipeline, reducing cloud spending costs by 25%. Pipeline implemented in Java Apache Beam on Google Cloud Platform's Dataflow.

• Increased observability of production systems by adding distributed tracing, cutting mean time to resolution of production issues from 1 day to 1 hour. Tracing added into a Go GraphQL backend service on Google Kubernetes Engine, through our Python machine-learning Google Cloud Functions, and exported to Honeycomb.

• Enabled anomaly detection and investigation of time series data from manufacturing lines. Increased manufacturing process engineers' same-week engagement by 20%. Analysis implemented in Python with SciPy, UI implemented in Highcharts and React with a GraphQL Go backend, microservice interaction over gRPC, with persistence in Google Cloud SQL and Heroic time series DB.

• Introduced ensemble programming (mob programming) and pair programming to the team, reducing pull request (PR) wait time from 4+ days to 4 hours, and eliminating 50% of PRs.

Developed high‐throughput annotation pipelines and report automation of patients’ genetic testing results.

• Put the release of a $75M USD investment in non-invasive prenatal genetic screening (NIPS) back on schedule by delivering critical features in our variant annotation pipeline. The NIPS product served 50+ unborn patients and their families per day. Pipeline implemented in Python (Django, SciPy) with a Postgres database.

Produced business intelligence software for Home Instead Senior Care.

• Replaced two internal services with serverless technologies, saving $12,000 in cloud spending and 120 hours of support work per year. Azure Cloud Functions implemented in .NET Core (C#) with Azure Cosmos DB.

• Delivered a service synchronizing 100+ daily client leads to internal CRM system. The service had zero known or reported bugs and outages during the 3+ years of its lifetime. Service implemented in Python with a SQL Server database.

• Led adoption of ensemble programming (mob programming), leading to an 8X reduction in change failure rate while maintaining rate of delivery.