Some work is built. Some is rescued. Some is verified. Some is still becoming. The status matters.
Problem: Real-life reflection, Scripture, and practical action live in separate places. Delivered: A privacy-first AI Christian life-intelligence system with consented connectors, user-reviewable observations, Bible search, plans, journals, and reminders.
Problem: Payroll publishing and employee access need a controlled operating layer. Delivered: Authentication, payroll import validation, payslip archive, protected PDF access, employee linking, announcements, admin views, and deployment environments.
Problem: A healthcare product needed confidence across API, mobile, identity, and release paths. Delivered: QA control-plane work, API execution, mobile testing, identity-boundary analysis, regression automation, and release-readiness reporting.
Problem: A single React Native codebase must produce controlled tenant variants without drift. Delivered: Manifest-driven analysis, conversion, sync, audit, doctor, and handoff workflows for Expo and bare React Native apps.
Problem: Wax deposition is not directly measured in many operating wells and pipelines. Research: A physics-informed Bayesian virtual sensor for latent wax state, restriction severity, location, growth, uncertainty, and abstention.
Problem: Students need one system for learning, revision, tests, exam preparation, and guardian clarity. Programme: A language-agnostic AI study platform with a named AI teacher, member-scoped access, and guardian accountability.
Problem: LLM workflows lose trust when sources, claims, and contradictions disappear. Delivered: A provenance-first knowledge CLI and Obsidian vault workflow built around immutable raw evidence and explicit source records.
Problem: Personal financial data, coaching, and lending decisions are fragmented. Direction: A financial intelligence layer that separates facts, behavioural signals, hypotheses, and predictions from credit decisions.
Attribution note: client names, logos, screenshots, and testimonials require permission. Proposal and research items are labelled by status.
There is code, a product workspace, or a public package to inspect.
The work entered an existing product and improved a meaningful workflow or system boundary.
Test execution, reports, or other evidence shows what changed and how confident we should be.
The question is real and the work is active, but the product is not being presented as deployed.
When the status is clear, the next decision becomes easier.
molaidrisLabs / evidence-led delivery