Whoopie Wanja

Led the end-to-end build of a multilingual knowledge-graph + RAG architecture for low-resource African language corpora, from data collection/cleaning through model evaluation and iteration. Integrated WHO/MOH structured health data with community-language text and fed evaluation results back into dataset improvements.

Supported Big Data labs and recitations on Spark, HDFS, and MapReduce for graduate students building large-scale data-processing pipelines. Debugged and optimized student Spark jobs and queries to improve pipeline execution reliability and performance at scale.

Built Python data pipelines for automated trend detection on a 25M+ user national health platform, contributing to improved system reliability. Presented model findings and data-driven recommendations to clinical stakeholders and delivered data literacy training that increased system adoption.

Built an NLP pipeline using spaCy and Hugging Face Transformers for medical prescription and information extraction. Contributed to model selection, annotation strategy, and evaluation with a distributed AI team.

Mentored learners in data skills, supporting structured curriculum delivery and technical problem-solving across a cohort.

Applied machine learning and analytics fundamentals in an African AI/education context, building the model-training and evaluation foundation for later NLP and data-engineering work.

Activities and societies: Deep learning, agentic AI, and ML in production coursework; graduate research focused on multilingual NLP and RAG.

Pursuing an M.S. in Engineering Artificial Intelligence, focusing on deep learning, agentic AI, and ML in production. Current graduate research assistant building multilingual NLP and RAG systems for low-resource African languages.

Earned a B.S. in Applied Computer Technology, establishing a foundation in ML and analytics used in later NLP and data-engineering work.