• Developed a hybrid machine learning model to forecast overall investor portfolio returns over the next 30 days.• Created a Bayesian recommendation system from scratch to suggest real estate projects for users.• Designed ChatGPT prompts for real estate ad generation using 25 marketing strategies in Arabic and English.• Built and deployed multiple APIs for various projects, using Docker and CI/CD pipelines for testing and deployment.

• Developed a deep learning model to classify JavaScript regular and obfuscated files, achieving a 97% accuracy.• Used deep learning to create a model that classifies JavaScript files as Web3 files, demonstrating an 83% accuracy.• Integrated MLOps principles to automate project workflows, from data preprocessing to model deployment.• Succeeded to build an automated stocks forecasting tool using machine learning and statistical models.• Conducted extensive research and implemented numerous proof-of-concept models to enhance forecasting accuracy.• Enhanced stock movement prediction accuracy to over 60% across various stocks using research-driven improvements.• Attained a 12.5% average revenue gain in the Turkish market and 4% in the US market in live daily trading testing.• Engineered a user-friendly car preference filter system, enhancing the car search experience.• Created a scoring framework to assess car conditions, aiding buyers through quantified evaluations.• Engineered a classifier model achieving a 99% accuracy, categorizing cars into regular-priced and high-priced.• Developed 2 specialized models for regular and high-priced cars, each boasting a 96% R-squared score.• integrated all three models into a unified predictive system, achieving high prediction performance.• Deployed the machine learning models into production and created an API to serve Turkish buyers and sellers.

• Built a predictive model for BigMart’s sales data of 1559 products across 10 stores, to find factors that influence sales.• Contributed to a team effort in developing a hit song prediction model based on repeated chorus patterns.• Attained a significant 67% accuracy by training the predictive model and integrated it with the frontend.• Built the backend system encompassing data processing, feature extraction, and model predictions.

• Implemented a linear regression model to predict student performance based on study hours with an MSE of 4.18.• Utilized unsupervised learning to identify optimal clusters in the Iris dataset and visualized results.• Built and graphically presented a decision tree classifier with high accuracy, improving decision-making for clients.• Created a hybrid model for stock prediction using numerical and textual analysis, achieving an MAE of 0.427.