Łukasz Michalak

I am a final-year student at PJATK, specializing in Data Science with a focus on machine learning and data analysis in Python. I am passionate about using data to solve real-world problems.

• Python Developer, Future Collars, 07/01/21, 09/30/21
• Full Stack Development Course

Classification and detection of brain tumor - master's thesis 

• Developed a deep learning models using TensorFlow & Keras on ~ 7,000 labeled MRI brain scans.
• Engineered custom CNN architectures and leveraged transfer learning models, achieving weighted F1-score of 0.975 with my own architecture. Applied multiple regularization techniques such as EarlyStopping, BatchNormalization, data augmentation and Dropout to enhance model generalization and prevent overfitting.
• Implemented YOLO for tumor detection, reaching 0.96 weighted F1-score.
• Used Matplotlib & Seaborn for data visualization and model performance analysis.

Apartment Price Prediction 

• Scraped approximately 6,000 apartment listings from Otodom using Selenium, filtering based on user preferences like city, price range.
• Cleaned and processed data using Pandas by handling missing values through imputation or removal, based on data analysis.
• Built predictive models (XGBoost, Random Forest, Linear Regression) and evaluated performance with MAE & MSE, performing hyperparameter tuning to optimize results.
• Developed an interactive dashboard in Streamlit and presented insights in Power BI.

Car Price Prediction 

• Sourced and cleaned Kaggle dataset using Pandas to ensure data quality and consistency.
• Performed feature engineering by adding new columns to improve model performance.
• Trained models (Decision Trees, Random Forests, XGBoost) to predict car prices, applying hyperparameter tuning for performance optimization.
• Displayed data using Streamlit and Power BI for interactive visualization.

Cinema Database Project (PLSQL & T-SQL) 

• Designed a relational database with 8 tables to manage films, screenings, reservations, and customer data.
• Created stored procedures for automated data validation and screening cancellations.
• Implemented triggers to maintain data integrity and automate updates.