Projects

Scam Data Collection & Threat Intelligence

At IwazoLab, I am leading the creation of a community-driven dataset on SMS and mobile scams in Africa.
This project supports the development of AI-powered threat intelligence and scam detection tools for financial protection. —

Led the creation of a social media listening and sentiment analysis tool that provides brand tracking and social insights to content creators and brands in Kenya. The tool supports informed decision-making in Kenya’s digital landscape and empowers local influencers with actionable insights.

Finlingo – AI-Powered Financial Literacy Platform

Founder of a conversational AI platform aimed at improving financial literacy and detecting mobile scams. Features include real-time scam detection, adaptive learning, and fraud awareness. Built in partnership with Afrela.

Uzazi AIoT – Maternal Health Monitoring Tool

Contributor to a maternal health initiative combining AI and IoT for affordable prenatal care. Integrates low-cost devices with intelligent monitoring for expectant mothers in low-resource settings.

Scam Dataset Collection & AI Detection System

Led a national-scale SMS scam dataset collection project. Presented findings at Deep Learning Indaba 2023.

Teaching & Leadership

NiweBora Leadership Program – Founder (2021–Present)
Mentored over 100 students transitioning from high school to university in leadership, civic responsibility, and emotional intelligence.

Data Carpentries Instructor – The Carpentries (Oct 2022–Present)
Taught 50+ learners in software tools, data wrangling, and reproducibility using Git, Python, and SQL.

Outreach, Media & Church Engagement

Sunday School Teacher & Church Social Media Strategy Team – ICC Kitengela

Media Engagements:
• Elevate TV – Panel on youth and innovation
• YouTube – Interview on social media analytics in Kenya
• Medium – Articles on AI, digital innovation, and civic tech

📂 Datasets

1. SMS Scam Detection Dataset – Kenya

Collected and annotated a real-world dataset of SMS messages from Kenyan mobile users, aimed at detecting high-risk and moderate-risk scam messages. The dataset supports research in fraud detection, trust in digital communication, and localized NLP.

Use Cases:
Supervised learning (Logistic Regression, XGBoost), risk classification, LLM robustness

Highlights:

Presented at Deep Learning Indaba 2023
Integrated into a Streamlit-powered demo
Includes metadata on message origin and scam typology

Access: Available upon request / In publication
Related Project: Finlingo, Scam Watch Initiative

2. Finlingo Dataset – Financial Literacy & Scam Education

Ongoing collection of multilingual financial literacy questions, scam typologies, and conversational intents used in the Finlingo AI-powered chatbot. The dataset enables adaptive learning paths and scam scenario training for underserved communities in Africa.

Use Cases:
Intent classification, multilingual NLP, AI tutors, financial education