Francis Onyango

Lead computer vision and forecasting projects end-to-end, from scoping with stakeholders through deployment and adoption. Built YOLO-based models for flower growth stage detection and yield forecasting, and designed multi-horizon forecasting models that cut average forecast error by 8%.

Built pipelines to pull and clean product data from major e-commerce platforms, processing 70+ products monthly. Trained models to estimate material composition, weight, and dimensions from listing data for life-cycle assessments.

Built customer segmentation and predictive models for marketing campaigns. Automated reports and built dashboards, saving an estimated 12 hours per week.

Built a computer vision app to grade cocoa beans by quality, reducing inconsistency in manual grading by 40%. Set up the full ML workflow on AWS using Ground Truth, SageMaker, Lambda, and API Gateway.

Built and deployed credit-risk scoring models for digital lending and pay-as-you-go financing. Built dashboards and a chatbot for health information access.

Built predictive tools for estimating relief package needs for disaster-affected populations. Used NLP and knowledge graphs to extract signal from messy sources for social-impact research.

Pursuing a Master of Science in Applied Statistics, with coursework and research in statistical methods and data analysis.

Completed a Bachelor of Science in Mathematics, building a strong foundation in mathematical theory and applications.