Hendersonville, North Carolina

Applied AI built aroundreal operations.

I identify operational problems, design the solution, direct AI coding agents, and test the result in the environment where it has to work.

How I work

I am a healthcare operator, clinician, and business owner with more than 20 years of experience. I am not a traditional software engineer, and I do not claim to have manually written every line of code in these projects.

My role is problem definition, product and workflow decisions, requirements, architecture and privacy judgment, agent direction, testing, troubleshooting, integration, deployment, and operation. AI coding agents implement software under my direction. Existing platforms, models, and services provide their own underlying capabilities.

Selected work

Systems shaped by the work itself.

Private AI system

Numa Private AI Platform

The problem

Create a useful AI assistant without treating privacy, access, and operating control as afterthoughts.

My decisions

I defined the use cases, requirements, privacy boundaries, model-routing approach, retrieval needs, memory, voice, tools, and monitoring. I also decided where local models and private infrastructure were appropriate.

How it was delivered

I directed Codex and Claude Code as the implementation team, then tested, troubleshot, deployed, and operated the resulting system. Existing models and services provide underlying capabilities.

Local models
Retrieval
Voice
Model routing
Monitoring

Numa ecosystem

One private system, several specialized capabilities.

Private coding agent

NumaCoder

Takes a project idea, breaks it into steps, writes and tests code, reviews its own work, and keeps progress safely checkpointed while building and maintaining real applications.

Computer-use assistant

NumaWorker

Carries out everyday computer tasks across files, email, research, apps, and websites. It performs the requested steps instead of only explaining how to do them.

Private service layer

Bridges

Connects Numa interfaces to models, memory, retrieval, voice, and approved tools while enforcing authentication, routing, and access boundaries.

Shadow-mode evaluation

Orchestrator

Proposes capability and routing decisions, records what it would have done, and remains deliberately separated from production authority until its evidence gates are met.

Vision and image workflows

Imaging

Supports controlled visual-input analysis and image workflows as part of the private Numa environment.

Healthcare workflow design

Clinic Operations & Touchpoints

The problem

Turn recurring front-desk and patient-retention problems into clear, manageable daily work.

My decisions

I designed workflows for new-patient follow-up, missed visits, retention, recurring-payment failures, cancellations, conversion tracking, and staff accountability.

How it was delivered

The work began with the operating reality: who needs to act, what information they need, which rules apply, and what must remain visible to management. I translated that into application requirements and directed development.

Workflow design
Business rules
Patient operations
Accountability

Business systems

Financial Operations Tools

The problem

Replace fragmented spreadsheets and manual financial tracking with clearer, auditable workflows.

My decisions

I defined requirements for transaction categorization, reporting, drill-down views, reimbursement tracking, mileage, trips, receipts, and audit history.

How it was delivered

I directed AI coding agents through implementation and revision, validated the tools against real transactions and operating scenarios, and corrected requirements when the software did not match the actual work.

Money Manager
Reimbursements
Reporting
Validation

Mobile and self-hosted system

Trail Maps & Private Navigation

The problem

Support private location history and route planning in real travel conditions, including unreliable connectivity.

My decisions

I defined the mobile workflow, background tracking behavior, offline collection, authentication, location history, mapping, self-hosted routing, and route-planning requirements.

How it was delivered

I directed implementation, installed builds on physical Android phones, and tested while walking and driving. Real-device failures drove further troubleshooting and iteration.

Android
Offline use
Self-hosting
Real-device testing

Approach

Direct, test, correct, repeat.

Find the real problem

Start with the operating environment, the people doing the work, and what repeatedly breaks.

Define the rules

Turn the problem into requirements, workflows, decision logic, privacy boundaries, and acceptance criteria.

Direct implementation

Give AI coding agents clear constraints, review their output, and redirect when the result misses the need.

Test in context

Use realistic data, edge cases, physical devices, and the actual operating workflow.

Deploy and improve

Integrate the system, monitor how it behaves, and keep responsibility for whether it works in practice.

Operating context

Technology informed by ownership.

20+years across healthcare, operations, and business ownership

4clinics led in a multi-location operating role

~13Ktreatments delivered across the clinics in 2023

2franchise businesses owned and operated

Healthcare and multi-location operations

Clinical care, staff leadership, operating procedures, performance management, patient retention, vendor relationships, and P&L responsibility.

Privacy-aware system design

Data access, private infrastructure, human review, and appropriate boundaries are treated as design requirements from the beginning.

Additional builds

Mobile AI

Field identification app

Practical workflow

Camping list app

Consumer concepts

Two tarot phone apps

Working prototype

Spa management and booking

These smaller projects are included as evidence of range. The primary case studies above are the clearest examples of my operating method and role.