$ls ~/projects/  — 5 items
chaguo-bright/ CIVIC TECH • ACTIVE

Chaguo Bright

Civic Education Platform

$cat README.md

Co-developing a multilingual civic education platform to improve access to leadership accountability information in Kenya, with planned regional expansion. Serves as a structured public ledger of leadership decisions, attaching accountability to named individuals over time.

03/2024 – Present Kenya → East Africa

$cat tech-stack.txt

AI Fact-Checker Multilingual NLP Python ML Civic Data
acki-drive/ SOCIAL IMPACT

African Children's Kick Initiative

ACKI Drive — Orphanage Support Platform

$cat README.md

A digital awareness and resource platform for orphanages across Kenya and East Africa, aimed at increasing visibility, funding, and adoption rates. Applies ML-driven engagement targeting to identify likely donors and volunteers, and explores recommendation systems for matching children with prospective foster families.

08/2024 East Africa

$cat tech-stack.txt

ML Targeting Recommendation Systems Python Web Platform
fraud-detection-ml/ RESEARCH • ACTIVE

Internal Fraud Detection

Machine Learning & Data Mining

$cat README.md

Ongoing research investigating ML and data mining techniques to detect and prevent internal procurement fraud, with focus on East African public sector contexts. Built and evaluated a baseline detection model; found ensemble approaches in human-assisted settings yield more reliable results than individual techniques.

01/2022 – Present Research

$cat tech-stack.txt

Python Scikit-learn Data Mining Ensemble Learning Outlier Detection
speech-emotion-recognition/ DEEP LEARNING

Speech Emotion Recognition

Deep Learning Model

$cat README.md

Designed and implemented a deep learning model to identify emotional states within speech audio, proposed as a decision-support tool for emergency responders and therapists operating without direct contact. Trained on the RAVDESS labelled audio dataset.

09/2021 – 12/2021 RAVDESS Dataset

$cat tech-stack.txt

Python Deep Learning Audio Classification TensorFlow
malaria-ml-distribution/ PUBLIC HEALTH

African Malaria Analysis

Resource Distribution via Machine Learning

$cat README.md

Developed an ML model to predict malaria hotspots across Sub-Saharan Africa and optimise the distribution of anti-malaria resources to high-risk regions. Sourced regional malaria index data from the World Bank. Achieved 79% baseline accuracy, with 88% projected after hyperparameter tuning.

01/2021 – 04/2021 79% baseline accuracy World Bank data

$cat tech-stack.txt

Python Scikit-learn Predictive Modelling World Bank API