Simble (Sreda Solutions)

I have built a data life cycle: first preprocessing of mobile sensor data, uploading it to cloud services (AWS), segmentation of driving, road features extraction, driving style clustering and risk prediction of a car accident.

NNFormat

Time series prediction of electric energy consumption of the factories-consumers and pricing on European electricity exchanges. Responsibilities: implementation of predictive models based on Statistical and Machine Learning methods. Predicted results allowed to save ∼600.000 kWh per month.

Lanit-Tercom

Customer churn prediction for mobile operator. Responsibilities: data cleaning and preparation, implementation of Principal Component Analysis, Random Forest and Decision Tree methods (programming in Python and R). Achieved results improved the determination of customer churn increasing the precision from 72% to 80% and were applied in customer retention policy by the mobile operator. Building recognition. Responsibilities: building recognition based on the face recognition algorithm with the use of Smart Augmentation in the computer vision project (Python).