Projects — Yosia Lukumai

01 both 2025

Free Space Optics

Laser communication for last-mile connectivity

Most communication infrastructure relies on fiber optics — fast, but expensive and difficult to deploy in remote areas. Free-space optics (FSO) transmits data through open air using lasers, offering a low-cost, high-speed alternative. We designed and built a complete FSO system from off-the-shelf components, including a custom beam-tracking mechanism to maintain alignment between transmitter and receiver. The system achieved data rates up to 120 kbps with very low power consumption — making it a practical option for rural and underserved communities.

PythonPCB DesignLaser OpticsHigh-Speed DesignCAD

02 both 2026 Active

Eleguard

IoT system for human-wildlife conflict prevention

Human-wildlife conflict threatens both farming communities and conservation efforts across Africa. Eleguard is a system of low-power IoT devices and a real-time monitoring dashboard designed to detect wildlife intrusions and instantly alert farmers and local authorities. Built on LoRa for long-range, low-power communication, the devices share a mesh of gateways and feed into a cloud dashboard. We piloted the system over three weeks in the field — with measurable impact — and began producing units for local farmers in partnership with MIT D-Lab and Twende Makerspace.

TypeScriptPCB DesignLoRaCADNode.js

03 software 2025

Poultry Disease Detector

Computer vision for early chicken disease diagnosis

A computer vision model for detecting three common poultry diseases — coccidiosis, salmonella, and Newcastle disease — from fecal samples. I collected and labeled a dataset of 500+ images, then trained a YOLOv8 classifier achieving 92% accuracy across all three categories. The solution includes a React Native mobile app backed by a Flask inference API, enabling farmers to photograph samples and receive real-time diagnoses directly from the field.

PythonYOLOv8React NativeFlaskComputer Vision

GitHub ↗

04 software 2025

Positive Cooker CRM

Mobile CRM for a clean-cooking social enterprise

A full-featured CRM mobile app built for Positive Ayin Company, a social enterprise providing clean cooking solutions to low-income households in Tanzania. The app enables the team to manage customers, track sales and inventory, and handle support — all from their phones. Built with Flutter and a Node.js/MongoDB backend, deployed on DigitalOcean. The company reported measurable gains in operational efficiency and revenue within the first year of use.

FlutterNode.jsMongoDBDigitalOceanREST API