About Foundations AI

I work with companies to build a strong AI foundation, grounded in best practices, proven methods, and solutions shaped around what your business actually needs. That means judging what I build by real performance and efficiency gains for your business, and staying current with emerging AI technology and evolving regulations so what I build holds up over time.

AI should make the people using it better at their jobs, not replace them. I build tools that handle the repetitive work and routine decisions, so people can focus on the parts of their job that need human insight. Data is handled with integrity, no shortcuts on how it's collected, used, or protected.

I partner with companies of all sizes, with specific expertise working alongside startups and small businesses.

If you're thinking about where AI fits into your company, I'd welcome the conversation.
Get in touch →

A closer look at the work behind Foundations AI

  • Built multi-agent systems, developed applied models, and designed knowledge graphs using Python, PyTorch, and TensorFlow on AWS infrastructure including Bedrock, SageMaker, Step Functions, and Neptune. Selected pipelines and tooling to match project and business needs.

  • Elevate Health Technologies — Artificial Intelligence Engineer (Contract)
    Feb 2025 – Aug 2026

    Architected and deployed a multi-agent AI system on AWS Bedrock, starting with a Prep Agent that dynamically routed payloads to downstream agents. Built a Test Claim Agent to automate input validation through Redox, eligibility checks through Cervey, and test claim execution against a live claims API, and a Rules Engine Agent that retrieved and created programs, workflows, and assessments with structured JSON output, cutting estimated workflow completion time by up to 75%.

    Built the Medication Alternative Recommendation Engine (MARE), which performed NDC lookups, queried a knowledge graph for drug alternatives, and ran inventory checks, along with a Clinical Transcription Agent that processed clinical audio into transcripts, medical summaries, and structured FHIR documents. That project secured an AWS innovation grant and saved the company roughly $125K in development costs.

    Orchestrated multi-agent workflows using AWS Step Functions, coordinating handoffs and shared state across autonomous pipelines. Designed and built a knowledge graph in AWS Neptune integrating patient, product, and clinic data to power drug alternative recommendations and diagnosis-based product suggestions.

    Xonar Technology Inc. — Artificial Intelligence Engineer
    May 2020 – Jan 2025

    Developed a novel multi-sensor CNN architecture for concealed weapon detection, reducing false positives by 40% and increasing true positives by 15%, deployed across multiple locations in the US. Built an object recognition model for x-ray imaging that reached over 95% detection accuracy with under 5% false alarms, and an end-to-end PyTorch pipeline that integrated directly with the company's existing tools and databases.

    Led a team of training coordinators and data analysts overseeing the weekly collection of tens of thousands of data points, and drove weekly discussions with the operating committee to keep everyone aligned on AI progress and challenges.

  • I. Tynes and S. Canavan, "Real-time Ubiquitous Pain Recognition," ACIIW, 2021.

    Guest Speaker, ASIS Chapter 30 Summer Seminar, 2024 — "AI: The Global Impact on the Security Industry"