At Tensium, I build AI and data pipelines and reinforcement learning environments with clear outcomes and grading criteria. I test tasks with Harbor and Claude models, then refine their wording, tests, and setup before review.
At Turing, I reviewed GitHub issues for realistic coding tasks used in LLM evaluation and AI training. I reproduced bugs and documented the steps needed to assess task clarity and correctness.
My project work includes a multimodal RAG proof of concept for searching structured and unstructured PDFs, combining retrieval and reranking to produce answers with citations. I also contributed to AI development for Nuru Women, an open-source women’s health platform that won Best Nostr Project at Hack4Freedom Nairobi 2026.
Across research and analytics projects, I’ve used Python, SQL, and Excel to turn data into analysis and reports. My work has included financial research, dashboards, and predictive modeling.
