I build practical artificial intelligence tools for fundraising, grant strategy, compliance, digital health, public affairs, and evidence-based decision support.

This portfolio is evidence for the Built By User idea: people close to a problem can turn their judgment into a working prototype, test the useful part quickly, and make the method easier to inspect through public code, source links, and explicit human-review boundaries.

Many of these tools are open for inspection in my public GitHub repositories. The code is not a claim that every prototype is production-ready. It is a record of how each tool works, what assumptions it makes, and where human judgment still belongs.

Federal Acquisition and Compliance

• ClauseFinder
  Searches FAR, DFARS, DAFFARS, eCFR snapshots, proposed rules, and deviation material for candidate clauses, prescriptions, source history, and applicability signals. It supports research and review. It does not issue a compliance verdict. Source code.
• Proposal Compliance Checker
  Compares a proposal draft with a solicitation and flags missing, non-compliant, and human-review items before submission. It checks completeness and compliance, not proposal quality or funding probability. Source code.
• NIH Subaward Checker
  Turns a specific NIH prior-approval notice into a practical grants-office check for domestic subaward changes. Deterministic policy logic remains separate from AI-assisted extraction and drafting. Source code.
• Nebraska Pork Compliance Copilot
  Provides evidence-constrained compliance decision support for Nebraska pork producers. It exposes sources and uncertainty instead of presenting generated text as legal or veterinary advice. Source code.

Fundraising and Grant Intelligence

• Funder Discovery
  Helps nonprofit teams discover possible funders, score fit against an organization profile, review public evidence, and avoid prospects that look attractive in a raw search but weak in practice. Source code.
• Funder Finder
  Turns a focus area or proposal draft into linked federal and foundation opportunities with cited fit signals. It is meant for a first prospecting pass, when the question is where staff time merits deeper research. Source code.
• Grant Fit Auditor
  Reviews a grant opportunity against a nonprofit profile, extracts requirements, ranks gaps, and recommends whether to pursue, pause, or decline. Its value rests on helping teams reject bad-fit opportunities before writing begins. Source code.
• Grant Decider
  A static companion implementation of the grant pursuit decision interface. It gives nonprofit teams a first-pass pursue, pause, or decline decision before staff invest in an application. Source code.
• Grant Analyst
  Performs evidence-first pre-submission review of eligibility, proposal merit, funder fit, evidence quality, and competitive readiness. It does not predict whether a proposal will receive funding. Source code.
• Foundation Grant Signal Lab
  Turns a proposal, applicant profile, and foundation strategy into a first-pass decision memo with visible evidence, review routes, funder risks, and next diligence questions. Source code.
• Grant Guardian Financial Analyzer
  Reviews nonprofit financial risk signals so finance, program, and fundraising teams can ask sharper questions before a grant advances. It is a due-diligence aid, not an automated approval system.
• Funder Discovery Plugin
  A Codex plugin and Custom GPT Actions pilot for nonprofit funder discovery, fit scoring, 990-backed funder briefs, donor briefs, and CRM-ready pipeline exports. Source code.
• Funding Fit Scorer Template
  A customizable, staff-only application template for scoring funding opportunities against an organization strategy and producing a reviewable go or no-go memo. Source code.

AI Workflow and Knowledge Tools

• Claude Skills Creator
  Helps teams turn repeatable expertise into structured AI workflows for grant research, compliance review, credibility checks, and funder-specific writing. Source code.
• Evidence to Audio
  A transparent multi-agent pipeline that turns a source pack into a private spoken-word brief while preserving inspectable intermediate artifacts. Source code.
• Network Radar
  Local-first relationship search for identifying people who can make introductions, referrals, and shortlists. Contact data, account metadata, tokens, and the working database remain local by design. Source code.
• Kindora for ChatGPT MCP
  A read-only Python MCP server that exposes Kindora public funder and grant tools inside ChatGPT for grant discovery workflows. Source code.
• NGO Executive OS
  An experimental AI executive-assistant starter kit for NGO fundraising and strategy, organized as reusable workflows rather than a single public application. Source code.

Healthcare, Public Affairs, and Strategy

• AI Under the Contract
  Models how one AI documentation product behaves under fee-for-service, MSSP, Medicare Advantage, and employer-direct contracts so a digital-health team can test incentives before choosing a business model. Source code.
• Healthcare Testimony Strategy
  Compares draft healthcare testimony with cited Senate committee records, identifies political risk and likely questions, and proposes safer answer frames for public-affairs review. Source code.
• Committee Analysis
  Analyzes congressional committee context so testimony, advocacy, and public-interest messaging can be prepared for the actual audience rather than a generic policy persona.
• Collins Public Record RAG
  A local citation-first research system for Senator Susan Collins’s public record. It answers from retrieved evidence and preserves source records for review.
• Senator Profiling System
  An evidence-first system for reusable senator profiles and cited testimony guidance. It labels observed facts, inferences, and recommendations so an analyst can review each layer.

Digital Health and Caregiver Tools

• First Signs Health
  Turns published cognitive-screening protocols into a family-usable browser workflow and a clinician-ready handoff document. It does not diagnose. Wayan.com version.
• Dementia Care and CCM Billing Guide
  Translates dementia-care and chronic-care-management billing choices into a practical decision aid for clinicians, caregivers, and families.
• Cognitive Screening Tool
  Combines clock-drawing and cognitive-screening workflows in a browser interface while leaving medical interpretation and follow-up with a clinician.
• Cog-Free Dementia Risk Estimator
  Explores a non-diagnostic risk-estimation workflow designed to help a user decide whether a clinical conversation merits attention.

AI Search, Education, and Public Tools

• GEO Audit Studio
  Creates an AI-search audit prompt, scorecard, measurement plan, and action pack so website owners can test whether AI systems can access, understand, cite, and accurately describe a site. Source code.
• College Fit Finder
  Makes college-ranking logic more inspectable by exposing the criteria and tradeoffs behind a recommendation instead of presenting a black-box list.

How I Build

Built By User is my build philosophy. A domain expert can test the useful core of an idea before committing to a long procurement or development cycle, then bring in engineering, security, legal, or clinical judgment where the work becomes sensitive or consequential.

I start with a concrete decision that wastes time or money: which funder to pursue, whether a proposal merits staff effort, what a compliance notice requires, which contract clause may apply, or what evidence should reach a clinician. I build the smallest useful tool around that decision, keep source material visible, and separate prototype value from production readiness.

Public code is part of the accountability model. It makes methods inspectable, limitations easier to find, and future changes easier to evaluate. Review the complete repository list at github.com/wayanvota.