Responsibilities include:

• Spearheaded a cutting-edge AI research paper submitted to CVPR, addressing data exploitation risks in deep learning with potential real-world impact, estimated to reach over 10,000 researchers and professionals.
• Designed a novel defense method using PyTorch, and diffusion autoencoders to secure DL models against data exploitation, processing 100,000+ data points to encode latent representations.
• Achieved a 60% reduction in model vulnerability in accuracy, significantly enhancing security and resilience against data extraction and reverse-engineering attacks such as adversarial training.
• Led and mentored 5 junior researchers, guiding their project execution and optimizing their code to ensure timely completion and high-quality outcomes.

Responsibilities include:

• Built an SQL-driven recommendation system using customer SMS and data usage patterns, leveraging a dynamic weighting mechanism to personalize GSM services, boosting engagement by 35% and conversions by 20%.
• Directed optimization and maintenance of a 3M+ record Oracle SQL database using Python with a 99.9% uptime, implementing automated workflows that saved 1,000+ operational hours annually. Implemented a distributed ETL pipeline using Apache Airflow and Python, optimizing data ingestion and transformation processes to handle 1TB of daily data with 30% faster processing times.
• Streamlined deployment with Jenkins and Docker, cutting time by 20% and downtime by 30%.

Responsibilities include:

• Led and built a 93%-accurate Turkish sentiment model (BERT, PyTorch) to detect rude retailers on an e-commerce platform, processing 50,000+ customer-retailer interactions and reducing complaint calls by 50% - generated $200,000 revenue for the company.
• Engineered a Python web scraper to automate customer complaint collection from 10+ sources, reducing manual data efforts by 60% and improving insights on production errors.
• Enhanced a chatbot with deep learning and IBM Watson, boosting user engagement by 20%, customer satisfaction by 15%, and adding $80,000 in projected revenue.

Responsibilities include:

• Developed a customer churn analysis web application leveraging to identify and retain at-risk clients. Enabled the bank to recover 10% of potential churners, resulting in improved retention for over 1,500 accounts.
• Designed interactive data visualization dashboards with Python (Dash/Plotly) to provide real-time insights into customer behavior trends, empowering stakeholders to make timely decisions.
• Conducted A/B testing on retention strategies, analyzing campaign effectiveness and recommending adjustments that increased customer engagement rates by 12%.