Consulting
Finance, grants, and AI systems.
I work with grant-funded organizations on budgets, reporting, and proposals. Through Sky Island AI, I also build voice systems and workflow automation.
Co-founder & CEO, Sky Island AI, Tucson, AZ
Tell me what you need
What I can help with
Fractional CFO & finance
I manage budgets, build reporting for boards, and tighten internal controls, including FAR and GAAP compliance. This is fractional CFO work for organizations that need senior finance help without a full-time hire.
Grant writing & management
I help find suitable federal, state, and local grants, write proposals, and manage the budget and reporting through renewal.
AI systems & automation
I build real-time voice interfaces, multi-agent systems, and workflow automation. The Virtual Case Manager and the development tools below are examples of this work.
I work on a retainer or by project, depending on the work.
Get in touchFinance & grant experience
At VisionQuest National, I managed the national budget across federal, state, and local funding. During four months running the development office, I wrote about 600 pages of proposals.
$70M
National budget managed at VisionQuest
$20M+
Annual recurring federal grants won
Selected work
Virtual Case Manager
AI case management with memory across sessions, for Medicaid, courts, and social services. Sky Island AI received a City of Tucson grant to deploy it for tenant support.
Python, Real-time Voice, Multi-agent
Economic impact dashboard for Tucson's proposed data center. Cited in national news coverage. 30+ media mentions and 10k+ views.
Data Modeling, Public Policy
Workflows
Coordinates agents in Claude Code and Codex CLI. Compares three models' independent reviews for high-stakes decisions.
Python, Multi-LLM Orchestration
single-shot-code
publicA nine-step code generation pipeline with reviews from three models, scope checks, and automated tests.
Shell, Multi-model Pipeline
medical-diagnostic
publicA diagnostic system that compares three models' assessments, checks for anchoring on an early answer, and includes human review.
Multi-model, Safety-critical
Prompt Testing Framework
publicTests prompts across model families using blind A/B comparisons, negative controls, and tests that remove one component at a time.
Prompt Engineering, Evaluation
proofer-extension
publicLocal-first AI proofreading for Chrome and Brave. Runs on Ollama, no data leaves your machine.
JavaScript, Ollama, Chrome Extension
Interactive data analysis of Arizona's 2024-25 fencing season. 1,596 competition instances, 398 fencers, 133 events.
Data Visualization, GitHub Pages
About me
I spent seven years at VisionQuest National, a grant-funded for-profit. As national budget manager, I reported to the CFO and CEO and worked directly with the Board chair. Before that, I ran materials and operations at a medical device company. My degree is from the University of Arizona, in economics and business administration, with an honors thesis in chemistry.
I've chaired the Arizona Division of USA Fencing since 2018, with around 500 members. I also chair the Still Point Fencing Foundation, which I set up as a 501(c)(3), and I coach there, running a staff of 11.
My first language was ASL. I have three deaf and hard-of-hearing sisters. I've traveled to 39 countries.
Contact
Email me with what you're working on and where you need help.
Contracts go through Sky Island AI.