PETER NDUNGU

I’m a Machine Learning Engineer specializing in end-to-end AI systems, designing, building, and deploying real-time inference solutions across NLP, deep learning, classification, regression, and anomaly detection. I focus on turning raw data into resilient, actionable business insights.

In production, I build low-latency FastAPI microservices (served via Uvicorn and Docker) and add rigorous data validation to ensure strict consistency between training and inference schemas. I implement MLOps foundations like CI/CD with GitHub Actions, automated model versioning and validation gates, MLflow experiment tracking, drift detection with PostgreSQL logging, and SHAP explainability—so models stay reliable after deployment.

My background combines applied analytics and operational data work: I’ve led structured dataset maintenance and KPI reporting as a researcher/analyst, and I’ve supported maintenance planning and performance optimization by analyzing SCADA operational data. I bring a careful, quality-first mindset from data cleaning through monitoring, with a goal of shipping systems that improve decisions and outcomes.