Data Scientist & ML Engineer
Sarah Ilyas
I build production ML systems that stay accurate over time — from drift detection and automated retraining to uncertainty quantification and model interpretability
PythonTime Series AnalysisPyTorchProphetSHAPOptunaMLflowFastAPIDockerAWSLangGraphStreamlit
Sarah Ilyas
Data Scientist & ML Engineer
Featured projects
View all →🌿
Cropcast
Production ML pipeline for agricultural yield forecasting across 13 countries. Automated drift detection, weekly retraining, FastAPI on AWS EC2.
xgboostmlopsfastapidocker
🥑
AvoGrade
Real-time avocado ripeness grading using ResNet18 transfer learning. Cache-first FastAPI serving with graceful fallback, batch scoring, and PSI/KS drift monitoring. 78% accuracy on a leakage-safe fruit-level split.
pytorchresnet18fastapidockerstreamlit