Software engineer and Director of Insights at Rwazi, based in Kigali, Rwanda. I build data and machine learning products for people on basic phones and thin connections, mostly in climate and health.
- Market Mosaic by Rwazi: a market intelligence data product I pitched and built from zero in January 2024. It had about 188,000 active subscribers in October 2026. The subscriber-scoring pipeline ranks readers for Sena, Rwazi's decision intelligence platform.
- Clisense: flood early warning for smallholder farmers on the Osun River corridor in Osogbo, Nigeria. Code: clisense.
- HealthDrive: telehealth over Unstructured Supplementary Service Data (USSD) for people without smartphones. The study is published as peer-reviewed research (DOI).
| Project | What it does | Stack |
|---|---|---|
| clisense | Benchmarks Random Forest, XGBoost, Decision Tree and an MLP for flood classification and serves the champion model | Python, scikit-learn, XGBoost, FastAPI |
| HealthDrive-Triage-Chatbot | Clinical triage chatbot: Llama-3-8B fine-tuned with QLoRA, BLEU 8.71 to 36.55 | Hugging Face, PEFT, Unsloth, Gradio |
| rwazi-sena-marketmosaic-pipeline | Pages through the beehiiv list, resumes after failures, retries on rate limits and scores readers for enterprise sales | Python, TypeScript, GitHub Actions |
| dengue-api-render | Predicts 2024 dengue cases for African countries from 2023 data | scikit-learn, FastAPI |
| climate-smart-ag-ml-project | Crop yield prediction from farm sensor data; tuned Random Forest reached R-squared 0.91 | scikit-learn, Keras |
| agbaje_rl_summative | Reinforcement learning agents (DQN, REINFORCE, PPO, A2C) in a custom farm-climate environment | Gymnasium, Pygame |
Agbaje, A. H. (2026). USSD-Based Digital Health in Rural Africa: A Machine Learning Research Direction for Low-Resource Health Signal Processing. International Journal of Innovative Science and Research Technology, 11(4), 3449 to 3460. doi.org/10.38124/ijisrt/26apr2523
Python, R, SQL, PostgreSQL, scikit-learn, XGBoost, TensorFlow and Keras, Hugging Face Transformers, QLoRA fine-tuning, FastAPI, Streamlit, Flutter, TypeScript, React, Power BI, Tableau.