Data Scientist
Current• Participating in the Aegean Region fire forecasting project carried out jointly with the General Directorate of Forestry by generating meaningful data using geospatial data, trying artificial data generation methods for the imbalance dataset, changing the loss function to improve performance, building and dockerizing the project pipeline.• Building credit risk model from scratch for Ziraat Bank POC.• Contributed to a demand forecasting project for TAT.• Rewriting the code in PySpark to make existing projects work on Sagemaker and Glue and making them triggerable with Lambda. Managing ECR and S3 containers.• Experimenting with feature engineering (selection and creation) using SHAP values for improving the explainability of models.