Retargeting Campaign Automation. Developed and deployed predictive models using CatBoost and logistic regression to score website visitors, automating the entire workflow from data preparation to segment deployment. Leveraged Docker and Airflow to integrate the process with marketing platforms, enhancing retargeting efficiency.Product Recommendation Model Optimization. Improved an existing on-site product recommendation model by retraining the LightFM algorithm and experimenting with advanced techniques, including fine-tuning large language models (LLMs). Collaborated with the MLOps team for deployment and validated its performance through A/B testing.Dynamic Landing Page Optimization. Developed and deployed a model to automatically route website visitors to the most effective product page, enhancing the likelihood of conversion. Managed data collection, feature engineering, and model development using CatBoost, followed by A/B testing to validate the model's impact, resulting in improved user engagement and conversion rates.Attribution Modeling for Marketing Optimization. Enhanced and implemented a machine learning-based attribution system to allocate marketing budgets more effectively across digital channels. Improved the model by using advanced methods alongside traditional approaches like first-touch, last-touch, linear models, and Markov chains to better identify the most impactful touchpoints.Technology Stack:- Machine Learning and Data Science Tools: Scikit-Learn, Pandas, Numpy, Optuna, XGBoost, CatBoost, - LightGBM, PyTorch, TensorFlow, LightFM- Data Management and Storage: MySQL, ClickHouse, S3- DevOps and Workflow Automation: Docker, Airflow, GitLab, Linux, Bash- Data Visualization and Monitoring: Superset- Development and Collaboration Tools: Python, SQL, JupyterLab, Visual Studio Code (VSCode)

Credit Scoring Model Development. Developed and integrated credit scoring models using alternative data from bank subsidiaries, applying Spark and Hive for data analysis and feature engineering. Trained models using LightAutoML and XGBoost, validating performance enhancements with logistic regression and preparing comprehensive reports for stakeholders.Trend Detection through Text Clustering. Applied advanced NLP techniques, including Latent Semantic Analysis (LSA), to extract hidden themes and patterns from external data sources. Utilized various clustering methods to group data based on content similarity, and created visualizations and reports to present insights and support strategic decision-making by leadership.Technology Stack:- Machine Learning and Data Science Tools: Pandas, Numpy, AutoML, CatBoost, Scikit-Learn, Logistic Regression- Big Data and Data Processing: Apache Spark, Hive, Hadoop, PySpark, HDFS- Development and Collaboration Tools: Python, SQL, JupyterLab