Kaan Gümele

Innovative and analytical data scientist with experience in deep learning, NLP, and computer vision. Published academic author with strong project management skills. Passionate about open-source development, real-time processing, and AI driven automation. Skilled in model development, performance optimization, and collaborative problem-solving.

Competitions & Achievements

• Kuika AI Hackathon | Finalist

• Led a team of three to develop an autonomous system that collects customer reviews.

• Enhanced the system with sentiment analysis and an improved rating mechanism, making it fully autonomous.

• Technologies: Python, LLM (Large Language Models), Natural Language Processing (NLP), Sentiment Analysis, Web Scraping, API Integration, SQL, Database Management

• Data & Analytics Challenge | Finalist

• Developed a revenue prediction model that forecasts 90-day revenue based on 15 days of gaming and metadata.

• Enabled data-driven strategic decision-making for gaming companies.

• Technologies: Python, PyTorch, TensorFlow, SciPy, Pandas, NumPy, Data Visualization (Matplotlib, Seaborn), Machine Learning, Time Series Forecasting, SQL

• GDZ Elektric Datathon | Finalist

• Gained experience in my first Kaggle competition and datathon by applying end-to-end data cleaning, feature engineering, and model development.

• Optimized models using boosting and tree-based algorithms.

• Technologies: Python, Boosting Algorithms, Tree-Based Algorithms, Data Cleaning, Feature Engineering, Model Optimization, Machine Learning, SQL

• Case Hunter | Finalist

• Performed web scraping and data processing through a custom-built API integration.

• Analyzed Migros and its competitors, integrating data science into market research for innovation.

• Earned a Data Analyst internship at Migros One as a result.

• Technologies: Python, API Integration, Data Processing, Data Preprocessing, Machine Learning, Market Analysis, SQL, Database Management

• Turkey Judo Championship 2020 | National Champion

• Became Turkey’s Judo Champion in 2020 after 13 years of competitive experience.

• Developed strong perseverance, teamwork, and strategic thinking skills throughout this journey.

References:

•Senior Data Analyst Beyza BEKAR |beyza.bekar@migrosonline.com • Asst. Prof. Muhammet Sinan BAŞARSLAN |muhammet.basarslan@medeniyet.edu.tr

Flutter İle Mobil Programlama, Udemy 

Etik Hacker Olma Kursu, Udemy

Oral Cancer Classification with CNN Based State-of-the-art Transfer Learning Methods

PROJECTS   

• Development of a Multi-Task AI Agent

Multi-agent Systems, LangGraph,GenAI, LLM, Pydantic  Designed and implemented a multi-task AI agent capable of managing multiple tasks simultaneously, making data-driven  decisions, and coordinating between task managers powered by Large Language Models (LLMs).  

• Document-Aware Chatbot with RAG and Vector Databases

RAG, LangChain, ChromaDB, LLM (Open Source), Streamlit  Built a chatbot capable of interacting with personal documents using Retrieval-Augmented Generation (RAG). Enabled  natural language conversations based on embedded vectorized knowledge from documents.  

• Entity-Based Sentiment Analysis from Real User Feedback

NLP, Web Scraping, NER, Sentiment Analysis, SQL  Performed entity extraction and sentiment analysis on Turkcell customer feedback. Provided actionable insights and  business recommendations through advanced LLM-based natural language processing pipelines.  

• Grade Detection of Meningioma Tumors (TÜBİTAK 2209-A)

Computer Vision, PyTorch, Deep Learning, CNN  Developed a clinical decision support system that detects tumor grade from 2D and 3D MRI images using CNN-based deep  learning models to assist physicians in diagnosis and treatment planning.  

• Revenue Prediction from Mobile Game Analytics

Time Series Forecasting, ARIMA, XGBoost, Machine Learning  Created predictive models to estimate 90-day revenue of mobile games using the first 15 days of gameplay and metadata.  Contributed to business intelligence and monetization strategy optimization.  

• Electricity Outage Prediction Based on Weather Data

Machine Learning, Pandas, Numpy, Scikit Learn  Predicted electricity outages at the district level using weather data. Enabled operational planning for workforce distribution  and reduced downtime through proactive maintenance strategies.  

• Autonomous Agricultural Monitoring with Mini Drones

  Python, OpenCV, ROS, Deep Learning, Computer Vision  Developed a fully autonomous mini-drone system for detecting plant diseases in strawberry crops using computer vision.  Delivered real-time visual feedback and statistical analysis to end users.  

• Segmentation-Based Detection of Dental Lesions

  CNN, Deep Learning, Computer Vision, Pytorch, UNET  Designed a custom segmentation model inspired by UNet using transfer learning backbones. Enabled accurate detection of  dental lesions from raw images, eliminating the need for dental dye, and enhancing diagnostic efficiency for dentists.