Euclide Mtisi

I'm a software engineer specializing in AI and machine learning, with a focus on the engineering behind models: pipelines, experiment tracking, and deployment. I build systems that are tuned deliberately, tracked rigorously, and deployed responsibly.

My recent work includes a telco churn prediction pipeline, where I tuned the decision threshold to 0.30 to cut missed churners (92.8% recall), tracked 30 Optuna trials and model versions in MLflow, and served the model through FastAPI and Docker for real-time inference. I also designed a closed-loop RFID medication dispenser with adherence verification, splitting control between a Raspberry Pi policy engine and an Arduino real-time controller to remove a single point of failure.

Alongside ML, I've shipped client-facing web applications and branding for South African clients using Python, JavaScript, React.js, Django, and SQL.