Professional Experience

Designed and built the ARMS Recommendation Engine, a room- and space-utilization system to maximize floorspace usage for healthcare and corporate-office clients, using Python, PostgreSQL, AWS, TensorFlow, and LLMs — built on Angus Remote Management System (ARMS) data and taken from proof-of-concept to a working product on a team of under 10.

• Created prototypes using LLMs, traditional ML models, and statistical approaches to optimize accuracy and usefulness — evaluating three technical directions (a conventional rules-based feature/candidate/policy pipeline, a TensorFlow model, and a GPT/LLM-based approach) before selecting the conventional pipeline to keep business rules hand-tunable, reserving ML for data-quality and anchor detection.
• Built an extensible, object-oriented plugin architecture where features, candidate builders, and policies are drop-in, auto-registering classes, enabling new data types, sensor types, and business rules without rearchitecting.
• Built and maintained four internal dev tools: a manual API test console, a read-only DB inspection console, a settings/config admin console, and a client-facing settings preview console.
• Designed the system for two verticals: healthcare, flagging booked-but-unused rooms as hospitals approach capacity, and corporate offices, driving meeting-room utilization, space repurposing, and capacity planning.

Designed tools and automations to improve SaaS deployments of IBM's OpenPages — an AI-driven Governance, Risk & Compliance (GRC) platform delivered as SaaS and built from microservices across multiple languages — in AWS and IBM Cloud, applying project management and development skills across Agile, Kubernetes, and cloud infrastructure principles. 16-month co-op on a ~2-dozen-person DevOps/SaaS team; shipped 100+ PRs and closed 50+ issues over the term.

• Developed automations and deployed worker node servers for CI/CD pipelines on Fyre VMs, reducing runtime failures by 80% across IBM Cloud's Platform Automation Toolchains, configured with Tekton and IBM's internal "One Pipeline" standard — writing and modifying pipeline scripts directly and resolving recurring pipeline-run and scanning-software issues, including auto-generated CVE GitHub issues.
• Owned the nightly promotion Slack reporting pipeline — a Python script run through a custom Tekton framework that posts CI/CD pipeline-and-code status to Slack — and kept it updated to surface the exact failing step as testing frameworks changed across microservices.
• Migrated the team's nonprod vulnerability-scanning pipeline (Twistlock scans via Auditree) from Travis CI to Tekton/IBM Cloud, including the pipeline's container image and its Python/Bash code, then triaged and resolved the resulting CVEs in microservice container images.
• Migrated pipeline-script-based jobs to Kubernetes Jobs as part of the team's move to ArgoCD and Helm, redesigning for Helm's stateless resource rendering.
• Led effort to modernize microservices and scripts to meet ARCH014 security/reliability standards & compliance, migrating microservices and images to Artifactory TaaS repositories — enabling package caching, controlled access, and JFrog Xray scanning.
• Worked under SOC2 and FedRAMP compliance requirements and GovCloud deployment context, covering all eight stages of the DevOps lifecycle (plan, code, build, test, release, deploy, operate, monitor).

Worked in the digital services team on healthcare and engineering-facility technical solutions, automating Autodesk Revit and AutoCAD workflows with pyRevit and custom Python/C# tooling.

• Built pyRevit-based tools and extensions for colleagues using Revit as a primary or secondary CAD tool, including a room-labelling tool that cut a roughly two-hour task down to under two minutes.
• Wrote Python and C# automation across Revit and AutoCAD to clean up visual elements, fix linked-document reference issues in multi-user Revit workflows, and handle other repetitive modeling tasks.
• Reduced export and cleanup time by 95%.