Work
Selected work.
Websites, models, maps and dashboards. Each one started with a real need and ended with something people could use.
Grant project · Water emergency alerts · 2026
SafeWatch.
People drown on the Volta Lake every year, and help often starts late because nobody knows where the accident is or who should go. SafeWatch sends an SOS with the phone's location to the person on duty for that stretch of water. If they don't answer in time, it calls the next person until someone does. Amadu Kamal designed and built it, from the phone apps to the dispatch screen and the server. It is in testing with the aim of a field trial at landing sites in the Volta Region.
- Works on any phone. An Android app, a browser app that installs from a link, SMS when there is no data, and a dial code menu for phones without internet.
- Nobody is left waiting. A named person must accept each alert. If they don't, it goes to the backup, then the coordinator, and also to the closest responder sharing their location.
- An alert when someone doesn't come back. Fishers say when they will be back. If they don't check in, the alert goes out by itself with their last position.
- Storm warnings in Ewe, Twi and English. The server checks the forecast every hour and drafts a warning for staff to send by SMS and phone notification.
- A record you can trust. Every step is logged with who did it and when, so response times can be reported to partners and funders.
- Built by
- Amadu Kamal, Model Analysis Hub
- Role
- System design, development, testing, deployment
- Stack
- Python, FastAPI, PostgreSQL, React, TypeScript, React Native, Leaflet maps, Africa's Talking, Render
- Year
- 2026
Get SafeWatch
On Android, open the downloaded file and allow the install when the phone asks. On iPhone, open it in Safari and choose Add to Home Screen. SafeWatch is in testing: in danger, also call 112. The free server can take a minute to wake up.
Screenshots use test data. See the SafeWatch front page.
Own product · Food delivery platform · 2026
Soma.
Soma is a local food delivery platform for Ho, Hohoe and the Volta Region. Customers order from vendors they know, riders deliver, and everyone is paid fairly and on time. Amadu Kamal designed, built and deployed it end to end through Model Analysis Hub, and it is now in testing ahead of launch.
- Four apps in one. A customer app, a vendor dashboard, a rider app and an admin console, each showing only what that person needs.
- Mobile Money and card payments. Paystack checkout, with the vendor and rider paid out automatically, and a flat GH₵2 service fee so everyone can see where the money goes.
- Notifications and an installable app. Order alerts in the app and on the phone, an installable web app that works on a weak connection, and delivery confirmed with a PIN.
- Built with security in mind. Strong password hashing, admin two-step sign-in, protection against forged requests and automated tests that cover the money flows.
- Built by
- Amadu Kamal, Model Analysis Hub
- Role
- Product design, development, payments, deployment
- Stack
- Node.js, Express, SQLite, JavaScript, Paystack, Leaflet maps, Render
- Year
- 2026
Screenshots use demo vendors and a test account.
Client project · NGO website · 2026
The Sunshine Project.
The Sunshine Project is a Ghanaian NGO that works with children, young people, women and vulnerable communities through health, education, mental well-being and leadership programmes. They needed a website that shows their work properly, and that they could keep current without a developer.
- Design and build. A warm, multi-page site: home, about, programmes, impact, opportunities, stories, get involved, contact, and a support page with Mobile Money and bank details.
- Content they control. An admin login where staff add programmes, change a programme from "launching soon" to a date to "completed", post schedules and upload photos. They can't accidentally break the design.
- Built to last. Works well on phones, loads quickly and is hosted for free.
- Client
- The Sunshine Project, Ghana
- Role
- Design, development, content system, launch
- Stack
- HTML, CSS, JavaScript, Python, Decap CMS, Vercel
- Year
- 2026
Proof it's mine
The site is live, and every page carries the credit in its footer:
Need a website like this for your organisation? Get in touch on WhatsApp.
Case study · Geospatial epidemiology · 2025
Malaria risk across Ghana.
An advanced spatial model that brings together climate variables, population density and historical case data to estimate malaria risk for each region of Ghana.
Why it matters
Malaria is one of Ghana's biggest health burdens, and resources are limited. Ranking regions by predicted risk helps decide where to send bed nets, rapid tests and treatment first.
- Data
- Climate, population density, historical incidence
- Tools
- Python, R, ArcGIS
- Year
- 2025
Risk of a high burden
Beyond a single ranking, the model estimates the probability that malaria prevalence in each region exceeds 30%. That turns a map into an early-warning tool: planners can see where the chance of a heavy caseload is high enough to prepare for.
Checking the model against reality
A model is only useful if its predictions hold up. Observed prevalence was compared with the model's predictions region by region, using residual diagnostics to see where the model agrees with the data and where it doesn't.
Dashboard
Healthcare utilisation dashboard.
A single screen that shows how patients use health services: average cost by diagnosis (ulcer, stroke, malaria, hypertension, diabetes), length of stay by risk category, and outcomes such as recovery, complications and deaths.
Filters for region and facility type (public or private) let a manager ask their own questions, like whether stroke cases cost more in one region than another, without waiting for a report.
Web platform · 2025
ProjectFlow.
A full-stack project management system for university research. It supports Undergraduate, Master's and PhD projects through the whole workflow, from draft submission to final publication.
Students track their work in one place, supervisors see where each project stands, and past work is searchable in a library of academic projects.
Product · 2026
QuantAI.
Model Analysis Hub's own product: a free research and statistics tool. It plans studies, picks the right test from assumption checks, fits regression models with diagnostics, forecasts time series, and exports every analysis as Stata, Python, R and SPSS code.
Try QuantAI