Aiden Ahmet Erdogan

- Industry generation system developed with PEFT LLM Llama & GCP, resulting in over 10% accuracy increase in ML matching models, reducing errors by 15%. - Operational efficiency boosted with 3 h/week saved in deployment time and 2 h/week in team collaboration through streamlined deployment processes and automated predictions in Slack. - Achieved a 35% ($32+K) reduction in monthly cloud expenditure via strategic optimization of virtual machines, ML models, and storage utilization.

- Enhanced sales prediction models, reducing errors by 10% through algorithm optimization, resulting in a $150K quarterly revenue boost. - Introduced dress combination recommendation system, driving a 5% sales increase, utilizing the Clustering method on AWS. - Developed a targeted customer segmentation model with 120 types, leading to a 30% campaign response rate rise. - Implemented Neural Machine Translation system for DeFacto websites, achieving a 95% accuracy boost, saving over $5K monthly.

- Secured $300 investment from foundations by optimizing chatbot multipurpose, introducing dialog engine, & real-time analysis dashboard, resulting in 20% higher engagement & 15% lower support costs. - Reduced CRF-based NLP chatbot's response time from 10+ seconds to under 3 seconds by Java to Python transition & algorithm updates. - Improved BERT NLP auto-labeling model, achieving an 85% acc. increase with TensorFlow, cutting labeling errors by 25%, saving 5 h/w, & optimizing workflows for cost efficiency.

- Reduced MAPE to <10% for Demand Forecasting, utilizing data cleansing, advanced feature engineering, and Light GBM with Linear Regression, resulting in 10% fewer stockouts and 5% less excess inventory. - Achieved F1 Score >0.9 in sentiment analysis on a 20 GB dataset, leading to 15% higher customer satisfaction and engagement via meticulous feature selection and Naive Bayes integration. - Collaborated on ETL execution and optimized 575 SQL queries for data extraction in an agile environment.

• Optimized data workflow, saving 10 weekly hours by creating a collaborative tool to split action data for the BA team, utilizing SQL and Python. • Performed data scraping from Twitter and sentiment analysis model in Python using Pandas and Scikit-Learn, pinpointing reviews relevant to users and yielding a 6% boost in sales. • Enhanced Company Secretary coaching, leading to a 26% reduction in customer complaints.