I am a data scientist who specializes in machine learning systems and LLM applications. I develop production-ready solutions across the full ML lifecycle model, development, and deployment with a focus on RAG pipelines and generative AI products. My work includes real-time intelligence platforms, automated content generation systems, and time series forecasting models. I work primarily with Python and cloud infrastructure to deliver measurable business impact through scalable ML solutions.

Please feel free to contact me on LinkedIn if you want to connect or work together.

• Designed and developed a fully automated system using LLMs to generate questions and answers from diverse input sources, including documents (PDF, DOCX), media (MP3, MP4), and online platforms (URLs, YouTube). Built an interactive web interface where users can test their knowledge with the generated questions and Receive a detailed performance report, showcasing end-to-end application development capability. https://github.com/yldzburhan/Digi-Did-I-Get-It-MultiModel-LLM-Projects
• FinSightAI is a financial news intelligence platform that helps analysts deal with too much information by turning a lot of market news into structured, ready-to-use outputs. It collects articles from many different places in real time, removes duplicates, and stores them. Then, it uses GPT-4 to write executive summaries and market bulletins. The system also has a RAG-based Q&A assistant that uses a ChromaDB vector store to give answers with source citations and time/source filtering so you can find them again. Users can export professional reports to PDF and see basic system activity on the dashboard. They can also see live market indicators like FX and global metrics. The platform generally simplifies daily market monitoring and enhances the reliability and speed of research workflows. https://github.com/yldzburhan/FinSight-AI
• MinutesFlow is an enterprise-ready meeting minutes generator that turns meeting recordings into reliable, reviewable outputs. It uses OpenAI models to transcribe audio and extract structured insights—action items, decisions, risks, and open questions—so nothing gets lost between discussion and execution. Built with Streamlit, MinutesFlow offers a clean, modern interface in a dark corporate theme and a review-first workflow that makes exporting and sharing minutes straightforward. https://github.com/yldzburhan/MinutesFlow
• DataBridge AI is a comprehensive Extract-Transform-Load (ETL) solution designed to address the challenges enterprises face when processing, validating, and integrating heterogeneous data sources. By combining deterministic data validation with AI-powered error correction, the platform significantly reduces manual data cleansing efforts while maintaining data integrity. https://github.com/yldzburhan/DataBridgeAI
• We created a step-by-step NLP pipeline to extract and analyse brand sentiment from unstructured Turkish texts. I led the work on a custom Named Entity Recognition (NER) model to capture brand mentions accurately and owned the full fine-tuning process for a pre-trained Turkish BERT model (T3 AI). Improving entity detection directly boosted the accuracy of the downstream sentiment model. The outcome was an automated tool that supports market research, competitive tracking, and real-time brand reputation monitoring. https://github.com/yldzburhan/NLP-Turkish-Brand-Entity-Sentiment-Recognizer
• We made a Python machine learning model that can guess how much carbon a person puts out each month based on things like their diet and how they get around. We made a simple "tree debt" metric that tells us how many trees we need to plant to make up for the damage that pollution does. This makes it easier to figure out what the results mean. Our bootcamp project won the "Best Graduation Project" award because it made it easier to find and more fun to look at environmental data. https://github.com/yldzburhan/CarbonFootPrintCalculator
• Developed an end-to-end machine learning project to predict customer churn for a bank. The process included comprehensive Exploratory Data Analysis (EDA), advanced feature engineering (TF-IDF, SVD for high-cardinality features), hyperparameter optimization with Optuna, and robust cross-validation. Various models like XGBoost, LightGBM, CatBoost, and a custom TensorFlow-based Neural Network were implemented within Scikit-learn pipelines, and the best performing model was deployed as an interactive Streamlit web application for real-time churn probability estimation. https://github.com/yldzburhan/bank-customer-churn-prediction-system

• Machine Learning Specialization - DeepLearning.AI
• Data Science & Machine Learning Bootcamp
• Machine Learning A-Z : Python,DeepLearning,AI
• Master Math by Coding in Python
• Master PostgreSQL