Proof point → Digital-skills programs need a delivery machine. I’ve built the platform such programmes run on - designed like a textbook, delivered like a product.

🩻 Problem

Digital-skills education in Nigeria is bottlenecked by delivery: courses live in WhatsApp groups and PDFs, with no enrollment lifecycle, no certification, no quality gate. Tricia Digital Academy needed a real platform for digital marketing education.

🔨 Solution

TDA - an e-learning backend built with unusual formality:

Architecture Overview

  1. Ten numbered design documents first - Scrum stories, requirements with UML, storyboards, use-case/activity/sequence diagrams, information hiding, class design and responsibilities, class hierarchies, collaborations, subsystems, coding conventions, and exception design.
  2. The specified product - four hierarchical roles, email verification, 2FA, a course lifecycle (draft → published → archived) with admin approval, cart checkout, and auto-generated PDF certificates.
  3. The implementation - NestJS 10 + Prisma migrations on PostgreSQL: auth module (JWT + local strategies, guards, roles decorator), courses, admin, email, and shared pagination/exception infrastructure - with co-located spec tests throughout (84 commits on develop).
  4. AI-orchestrated delivery - ~25 task branches (codex/implement-jwt-authentication-with-tests, codex/configure-ci/cd-pipeline-for-backend…) merged through a develop/staging/main branch ladder, governed by an AGENTS.md.
  5. Operations literacy - a companion tutor-setup repo scripting an Open edX-style LMS stack (CMS, MFE, MongoDB, Elasticsearch, MySQL, Caddy) in Docker.

Quality signals

CI workflow, Husky pre-commit/pre-push hooks, three docker-compose profiles (dev/test/prod), changelog, issue templates, code of conduct, Apache license - the strongest engineering-process discipline in the education portfolio.

📜 Philosophy

Education platforms deserve textbook engineering: requirements → UML → class design → conventions → exception policy, then small, test-gated increments. A deliberate, documented upgrade from my earlier schema-first style.

🎓 Key learnings

  • NestJS/Prisma/PostgreSQL after years of Mongoose - choosing relational integrity for enrollment and certification data.
  • Formal design methods applied to a real product, not a classroom exercise.
  • Orchestrating AI coding agents through per-task branches, review gates, and CI - a delivery model that multiplies a small team.
  • Operating existing open-source LMS stacks (Open edX/Tutor) rather than rebuilding them.

📈 Output & impact

  • A 10-document design suite plus an implemented, test-covered backend with CI/CD, delivered in ~2.5 weeks of dense work (July–August 2025).
  • A reusable blueprint: design documents an institution could audit, attached to code that passes its own gates.

🌍 Why this matters

Talent & Education. Training programs at any scale - bootcamps, academies, institutional skills initiatives - are an e-learning delivery problem: enrollment, certification, course approval, quality gates. This is that machine, documented to a standard an institution could inherit - plus the operations literacy to run open-source LMS stacks at near-zero licensing cost.

🚀 Hire me

Building an education platform or skills program that needs real delivery infrastructure? Let’s talk → · See also: E-Student · CS Foundations · The thesis