Context

Public health agencies in Maine needed a way to put community-level health estimates directly in front of the people making resource-allocation decisions — without funneling every question through an analyst. County-level dashboards existed; sub-county, subpopulation-aware tooling did not.

Challenge

Build a production platform that lets non-technical partners at MaineHealth and Portland Public Health explore Apriqot's estimates interactively — choosing datasets, geographic levels (census tract, township, ZIP), and subpopulation filters — then save, view, and export the resulting maps. The system had to handle authenticated multi-user state, render statewide choropleths fluidly, and stay legible under real partner workflows, not demo conditions.

Approach

Built the frontend in Next.js (App Router) with TypeScript, Tailwind, and shadcn/ui, with Leaflet driving the interactive map layer over OpenStreetMap tiles. Designed and implemented the data selection UI — dataset, geography, and subpopulation pickers — plus the saved-map persistence flow and tabular export. On the backend, wired authentication through Auth.js and modeled per-account map state in Postgres. Wrote unit and end-to-end test coverage in Jest across the estimate-retrieval and saved-map flows to keep the partner-facing surface stable across iterations.

Outcome

WHAM! is in active use by partner organizations in Maine, including MaineHealth and Portland Public Health, where it backs decisions on mobile food pantry placement and subpopulation-targeted outreach. The platform now serves as the delivery surface for Apriqot's full estimate catalog — extending past food insecurity into other community-health domains.

Context

Mother of Fact needed a tool that hospital administrators and clinic decision-makers could use themselves — not through a sales call — to see the financial and clinical case for adding maternal nutrition services to their care model. The audience was busy practice leaders who needed defensible, partner-ready numbers in a format they could hand to a CFO.

Challenge

Translate a multi-variable clinical and financial model — birth volume, payor mix, dietitian coverage, complication rates, and cost impact across cesarean, NICU, gestational diabetes, and preeclampsia — into a self-serve product that ends in a branded PDF report delivered to the user's inbox. The math had to be transparent and editable as clinical assumptions evolved, the input flow had to feel light, and the output had to look credible enough to circulate inside a partner organization.

Approach

Architected and built the system end-to-end as a Vite + React + TypeScript single-page app on AWS. Designed the deterministic ROI engine in TypeScript, computing future value, ROI percent, per-condition cost savings, RDN capacity, additional patients served, and three-year revenue projections from user inputs against editable clinical and financial assumptions. Wired the frontend to four AWS API Gateway endpoints for defaults retrieval, defaults persistence, submission logging, and transactional email through Lambda + SES. Built the reporting layer with Recharts for in-app visualization and a pdf-lib pipeline that overlays computed results onto a versioned S3-hosted report template, with @react-pdf/renderer as a programmatic fallback. Built an internal admin surface so the Mother of Fact team can update clinical and financial assumptions without code changes. Set up the static deployment pipeline to S3 + CloudFront with scripted build, sync, and cache invalidation.

Outcome

In production and publicly linked from Mother of Fact's provider site. Mother of Fact reports that partner clinics using the underlying nutrition care model see an average 49% return on investment, $55,920 net benefit, and 20% decrease in pregnancy complications — figures the tool surfaces and personalizes for each prospective partner. The estimator replaces what was previously a manual analyst workflow and now serves as a primary entry point for new clinical partnerships.