Why We Built a Software & AI Engineering Program for African Developers

I started McTaba Labs because I watched talented developers in Kenya and across East Africa finish bootcamps and online courses, then struggle to build anything useful for the market they lived in.

They could build a React app. They could set up a Node.js backend. They had GitHub repos full of projects. But ask them to integrate M-Pesa STK Push, or build a USSD menu for a feature phone user, or wire up WhatsApp Business API for order notifications, and they froze. Nobody had taught them.

The reason is simple. Almost every coding bootcamp and online course in the world was designed for the American market. The payment integration is Stripe. The messaging is Twilio. The assumption is that your users have iPhones, high-speed internet, and credit cards. That describes San Francisco. It does not describe most of Africa.

M-Pesa processes over $300 billion in transactions annually. More money moves through M-Pesa in Kenya than through the country's traditional banking system. WhatsApp has over 100 million active users in Africa. USSD still reaches more people in most African countries than any smartphone app.

These are not niche tools. They are the backbone of how African businesses collect payments, reach customers, and deliver services. Engineering with these systems requires specific knowledge:

• M-Pesa's Daraja API has its own authentication flow, callback patterns, and reconciliation requirements that work nothing like Stripe
• USSD sessions are stateless and timeout-sensitive; building good USSD requires a different UX discipline than web development
• WhatsApp Business API has rate limits, template approval processes, and messaging windows that shape what you can and cannot automate
• Airtel Money, MTN MoMo, and other mobile money systems each have their own APIs and quirks

This is real, specialized engineering knowledge. It cannot be picked up in a weekend tutorial because there are very few tutorials. The documentation is often incomplete. The best practices live in the heads of developers who have already built and shipped with these systems.

We decided that if no one else was going to build a serious curriculum around this, we would.

When we started seeing AI tools become genuinely useful for developers in 2024 and 2025, a new gap opened.

Bootcamps responded in two ways. Some ignored AI entirely and kept teaching the same curriculum they had been running for years. Others bolted on a "prompt engineering" module at the end of their program and called it AI-ready. Neither approach prepares you for the reality of building AI-powered products.

Worse, AI coding tools themselves have a bias problem. Ask Copilot or Claude to build you a payment system and they will generate Stripe code. Ask them to build a messaging integration and they will reach for Twilio. They do not know about M-Pesa's STK Push flow. They do not understand USSD session management. They will not suggest Africa's Talking as your telephony provider.

This means AI makes the developer who only knows generic, Western-stack skills more replaceable. AI can generate that code. But it makes the developer who knows the African Stack more valuable, because that knowledge is exactly what AI lacks.

We built context engineering, RAG, and agent development into the curriculum from Phase 2 onward. Not as a separate module, but as part of how students learn to build every project. The goal: graduate developers who can both use AI tools effectively and build AI features into the products they ship.

We believe the most valuable developer in Africa over the next decade is someone who can do three things:

1. Build full-stack web applications using modern tools (React, Node.js, PostgreSQL, the fundamentals that every market needs)
2. Integrate with the African Stack (M-Pesa, WhatsApp, USSD, mobile money) because that is where African businesses operate
3. Use AI as a multiplier, not a crutch (build AI-powered features, use AI tools to work faster, understand the capabilities and limits of LLMs)

That combination is rare. Most developers have one of these. Very few have all three. The Software & AI Engineering program exists to produce developers who have all three by the time they graduate.

We could be wrong. Maybe the African Stack stays niche. Maybe AI tools get so good that specialized knowledge stops mattering. We do not think so, but we hold our bets honestly.

What we are certain about: the developer who can build, ship, and integrate real systems for real African businesses will have no shortage of work. The demand for that skill set already outstrips supply, and AI is widening the gap, not closing it.

We are a small program. We run small cohorts (around 10 students) because that is what allows real mentorship. We do not have a campus. We are not accredited as a university. We cannot guarantee you a job.

Our graduate outcomes data is still being built. Where we have verified information, we share it. Where we do not, we say so explicitly rather than inventing numbers. We think honesty about limitations builds more trust than polished marketing that crumbles under scrutiny.

We also know that our program is not right for everyone. If you need an accredited credential, go to university. If you cannot commit 20 hours a week, try the self-paced Academy course. If you already have years of professional experience, our beginner-friendly pacing will frustrate you.

But if you are someone who wants to build products for the African market, wants AI engineering skills alongside your software development skills, and wants the structure of a mentored cohort with real deadlines, then this is what we built for you.

If what you have read here resonates, come see for yourself. Join the Discord and talk to current students and alumni. They will give you a more honest picture of the program than any marketing page. Read the full curriculum breakdown and decide if the projects and skills match what you need.

If they do, apply. If they do not, that is fine too. We would rather you find the right program than join the wrong one.