I'm Chewi.
I'm a final-year Software Engineering student at The ICT University in Yaoundé, Cameroon, passionate about machine learning and AI alongside full-stack and mobile development. I split my time between backend-heavy systems work (Django, DRF, message queues, microservices) and mobile development in Flutter and native Android. I'm most interested in building things that solve a problem I can see in front of me, rather than optimizing a metric on a dashboard.
How I approach engineering
Most of what I build starts from a gap I notice directly: students with no free payment sandbox to build against, Cameroonian citizens with no accessible legal guidance, budget Android devices that off-the-shelf safety apps quietly ignore. I try to ship the real, unglamorous infrastructure underneath those problems, auth flows, async task queues, sensor fusion, vector search, not just the UI on top.
Ask before you build
Most wasted engineering time comes from building the wrong thing well. I default to understanding the real constraint first.
Design for longevity
Code that survives past the demo. Migrations, auth, and data models get the same care whether it's a class project or a startup.
Decompose until obvious
Microservices, background workers, and clear API boundaries: complexity should live in small, individually understandable pieces.
Build for real use
Simulated payment rails, bilingual chatbots, sensor-fusion safety apps: I'd rather ship something a real person in Cameroon can use than another CRUD tutorial.
The journey so far
I started at The ICT University without a clear idea of what kind of engineer I wanted to be, and coursework alone never quite scratched the itch. That changed once I started co-founding real projects with classmates. PayCam grew out of a simple complaint every student developer in Cameroon shares: there is no free, realistic sandbox to build payment features against. So I built one, wallet app and SDKs included.
LexCam and Ask-Jo followed a similar pattern, a real access gap (legal information, bilingual mental-health and career guidance) that a small team of us could actually close with the tools we were learning: Django microservices, RAG pipelines over ChromaDB and Qdrant, FastAPI services calling Llama 3.3 70B. Along the way I've also drifted toward mobile and embedded-adjacent work, Kotlin, sensor fusion, on-device fall detection, because the hardware constraints of budget Android devices common across Cameroon are a genuinely interesting engineering problem that most tutorials skip.
Outside of shipping code, I spend a fair amount of time reading about distributed systems and applied ML, and I'm slowly building an interest in IoT, mostly because half of what I want to build next (safety hardware, agri-tech sensors) eventually needs a physical device attached to it.