// Writing
From Firewalls to Foundation Models: Why I'm Betting on AI Engineering
From Firewalls to Foundation Models: Why I'm Betting on AI Engineering
By Olayinka Samuel Ojo — Founder, Lit Creative Designs | Abuja, Nigeria
I have a confession: I didn't start in AI.
I started in design. Over eight years of brand identity, print work, UI/UX — the kind of work where you're obsessing over kerning at 2am and arguing about whether a shade of blue is "too corporate." Then I moved into software engineering. Then cybersecurity. ISC2 CC. Network+. Cisco's NetAcad. 3MTT. I kept adding things, not because I was restless, but because each new discipline kept making the previous one make more sense.
And now I'm adding AI engineering.
Why now, and why this
There's a version of this story where I say I had a eureka moment. I didn't.
What I had was a slow accumulation of evidence that AI wasn't coming for my field — it was becoming my field. Every client conversation at Lit Creative was starting to include the words "can we use AI for this?" Every cybersecurity briefing I read had a new section on LLM vulnerabilities, prompt injection, and model poisoning. The software engineering curriculum I was grinding through had me thinking about how language models handle text transformation in ways that weren't entirely different from what I was building by hand.
At some point, you stop treating AI as an adjacent thing and start treating it as the main thing.
What I'm actually learning
I'm not starting from zero — and I think that matters more than people admit.
When you come to AI engineering with a background in networking, you already understand latency, throughput, and why distributed systems fail in annoying ways. When you come from cybersecurity, you think adversarially by default — which turns out to be exactly the right mental model for evaluating LLM safety. When you come from software engineering, you know that elegant demos and reliable systems are completely different things.
Right now I'm working through the foundations: Python, NumPy, Pandas, and fast.ai's Practical Deep Learning course. I'm also going deeper on the Anthropic API, which I've been building with for a while through my studio. The plan is structured — classical ML intuition first, then LLMs and RAG pipelines, then production deployment and MLOps, then a specialisation in AI security, where my cybersecurity background becomes a genuine edge rather than a footnote on a CV.
I've already shipped real products
This matters, so I'm going to say it plainly: I know what it takes to take something from a brief to a live product.
Through Lit Creative Designs, I've built and shipped production web products for clients across Nigeria — TutorNest, an education platform; TravelBank, a travel finance product; Coplan Associates, Nalado, Power ADF, and Manyatta, among others. I've handled brand identity for the NBA Annual General Conference 2024, produced conference materials for the NBA-SBL 20th International Business Law Conference, and built ComplyNG — a mobile-first compliance education game covering Nigerian business law — for a RegTech hackathon.
The AI work I'm building toward isn't theoretical. It's the next layer on top of real delivery experience.
The Nigeria angle
Here's something I don't see discussed enough in the global AI conversation: the problems worth solving in Nigeria are genuinely different, and genuinely interesting.
A compliance RAG tool that actually understands FCCPC guidelines, CBN regulations, and CAMA provisions. A scam classifier trained on Nigerian SMS fraud patterns — advance fee fraud doesn't look the same as phishing emails targeting UK banking customers. Document intelligence tools built for the realities of African SMEs, where contracts get scanned from photocopies and financial reports come in formats that no Silicon Valley dataset ever prepared a model for.
These aren't niche problems. They're underserved problems in a market that's growing fast. And the engineers best positioned to solve them are the ones already here, already building, already talking to the people who need them.
I'm building in Abuja. I'm part of the tech community here. The projects I'm going to build over the next 18 months are useful to people I can actually talk to and get feedback from. That's an advantage I'm not taking for granted.
What I'm building toward
The honest answer is: an AI engineer who can also think like a security researcher. Someone who can build the pipeline and stress-test it. Who can ship a product and understand what breaks when someone tries to abuse it.
That's not a common combination. I'm not claiming to have it yet. But I know what the path looks like, and I'm on it.
I'll be writing here regularly — what I'm learning, what I'm building, where I'm confused, and what finally clicks. If you're somewhere on a similar path, especially in Nigeria or across Africa, I'd genuinely like to hear from you.
The work starts now.
Olayinka Samuel Ojo is the founder and Creative Director of Lit Creative Designs, a multidisciplinary creative studio based in Abuja, Nigeria. He writes about AI engineering, cybersecurity, and building tech from Africa. You can see his work at litcreativedesigns.com and olayinka.name.ng.