Product
Projects, files, invoicing, messaging, and analytics demos in one application.
- Full project lifecycle: inquiry to delivery and payment
- Secure file upload, download, and access logging
- Quoting, invoicing, and payment tracking
How it works
How we build it, and how you can check.
We build health analytics for institutions, researchers, NGOs, and public-health teams across Africa. Here is how we design models, validate them, and report what they change.
MethodologyValidationGovernanceOutcomes
Calibration check
How we read a calibration plot: predicted probability against observed frequency, measured against perfect agreement. Illustrative.
Methodology
From question to validated model
Each step is documented, and the harder steps carry the most weight in how we judge whether a model is ready.
01Our approach
Start with the question, not the algorithm
We define what needs to be answered and choose inputs that exist in the data, then pick the method. The decision the output supports comes first.
02Validation
Test against real data
Validation matches the intended use and data. Advanced models must beat a simpler baseline to earn their place.
03Calibration and fairness
Read calibration and subgroup performance
We assess calibration and interpretability, not just headline accuracy, and review how models perform across different subgroups.
04Transparency and privacy
Document the limits
Limitations, assumptions, and risks are documented clearly. Models are built for real data, with privacy built in.
The proof, read directly
Performance is reported per subgroup, not as one number
A single accuracy figure can hide where a model fails. We hold each subgroup against a tolerance band and document the gaps so the people relying on the output know exactly what it covers.
Subgroup performance
Illustrative subgroup view. Real figures are reported per project.
What you can verify
Product, governance, and operations
Three tracks of evidence, each one you can check against what the platform actually does.
Governance
Scope, privacy, and usage boundaries are documented publicly and enforced in the product.
- Public scope and intended-use documentation
- Security and privacy statements that match what the platform does
- Demos use sample data only
Operations
Every significant action is logged. Admins see project history, enquiries, finances, and workload.
- Full audit trail for changes, assignments, and reviews
- Enquiry queue with delivery tracking
- Finance, workload, and customer dashboards
What it changes
Better decisions, not just better models
When teams can interpret data earlier and report outcomes clearly, they plan better with limited resources.
Monitor
Track programme performance against the same evidence base.
Report
Turn results into clear, structured outputs teams can act on.
Plan
Decide where to focus limited resources with clearer signals.
Work with us
Talk about a study, pilot, or analytics project
Bring a model to validate, a study to design, or data to interpret. We will tell you how we would approach it.