Full-stack engineer. Backend-first, production-first.

Aropet Joel

I build software that holds up under real conditions: C#, NestJS, React, Next.js, PostgreSQL, SQL Server, Linux. Precise where it counts, calm everywhere else.

Get in touch See the work

01 · About

About me

I'm Aropet Joel, a full-stack engineer based in Uganda, working backend-first: C#, NestJS, and React or Next.js on top, PostgreSQL or SQL Server underneath, deployed and run on Linux. I like owning a system end to end, from the schema to the API to what a user actually sees, but the backend is where I spend most of my time and where I do my best work.

That's also the reason behind KiboSystems, the company I'm building: too much of the software running everyday business in Uganda, and the region more broadly, is fragile, manual, or falls over the moment the connection drops. KiboSystems exists to fix that, infrastructure that's actually mastered, not patched together and hoped for.

If you're hiring for backend or full-stack work, or want to talk about what KiboSystems is building, I'd like to hear from you.

Technical expertise

Core technologies and tools I work with

C#
NestJS
React.js
Next.js
PostgreSQL
SQL Server
Linux

02 · Work

Projects

AI-Powered Offline SMS Chatbot
A chatbot that lets users interact with AI over SMS without an internet connection. Built with FastAPI and MongoDB, integrated with Africa's Talking for SMS and OpenAI models for responses, with Ngrok exposing the local server securely.
Problem
In regions with limited internet access, traditional AI chatbots are inaccessible. The goal was a solution that lets users reach AI capabilities via basic SMS, bridging the digital divide.
Role
Lead full-stack developer, responsible for system architecture, the FastAPI backend, SMS gateway integration, and AI model interaction.
More detail
Challenges
Integrating a third-party SMS API (Africa's Talking) with reliable delivery, managing real-time AI responses over a low-bandwidth channel, and securely exposing a local dev server via Ngrok for testing and deployment.
Impact
Lets someone with a basic feature phone and no data plan hold a real conversation with an AI model over plain SMS. The Africa's Talking gateway handles delivery, the FastAPI backend manages session state between texts, and an OpenAI/Deepseek model handles inference, so the internet connection requirement drops out of the equation entirely.
Lessons learned
Robust error handling matters more in distributed systems. Data transfer needs deliberate optimization for SMS. AI models can be adapted for efficient use on limited compute.
FastAPIMongoDBOpenAINgrokAfrica's TalkingDeepseek:r1-7b
Nyumbani: Home Rental Application
A modern home rental platform for seamless property discovery, filtering, and management. Built with Next.js on the frontend, FastAPI for a scalable backend, MongoDB for storage.
Problem
The existing home rental market lacked a user-friendly platform with real filtering and efficient management for both tenants and landlords.
Role
Full-stack developer, covering the Next.js frontend and FastAPI backend, with a focus on user experience and data integrity.
More detail
Challenges
Complex search and filtering logic, real-time property availability, and optimizing MongoDB queries at scale.
Impact
Replaces manual, listing-by-listing browsing with real filtering, by price, location, and availability, backed by MongoDB queries tuned for the access patterns tenants and landlords actually use.
Lessons learned
State management with Zustand in complex Next.js UIs. Designing efficient NoSQL schemas. The value of a microservices-oriented approach for scalability.
Next.jsTailwindZustandFastAPIMongoDB
Fish and Fish Products Marketplace
An online marketplace connecting fish vendors with buyers for a streamlined buying experience. Built with Next.js and MySQL.
Problem
Local fish markets lacked a centralized online platform, limiting market reach for small vendors and making it harder for buyers to find fresh product.
Role
Frontend developer, responsible for the UI and its integration with backend APIs for listings and user interaction.
More detail
Challenges
An intuitive, responsive UI for two different audiences, efficient image handling, and secure transaction flows.
Impact
Gives small fish vendors a storefront reachable outside their physical market radius: buyers can browse and contact vendors directly online instead of relying on in-person visits or word of mouth.
Lessons learned
Responsive design matters more for e-commerce than it first appears. Efficient fetching and caching in Next.js. Best practices for user-generated content.
Next.jsTailwind