Data Scientist( Globerce Inc. )

Описание

Tulegen Didar
R tulegendidar52@gmail.com |+7(747)-540-5138 |� linkedin.com/in/didar-tulegen-506a63308
� github.com/diiidar | kaggle.com/didartulegen
General Info I am actively developing in the field of machine learning and data analysis. I am looking for an intern or junior specialist
position in machine learning to apply data science skills, model building and continuous professional development.
Education Suleyman Demirel University
Almaty, KZ
Bachelor in Computer Science Sep. 2022 – May 2026 •Relevant Coursework: Advanced Algorithms, Database Management Systems, Machine Learning, Deep Learning,
Computer Vision.
Experience Independent Machine Learning Pro jects (Kaggle)
Oct. 2024 - Ongoing
• Applied ML techniques to real-world datasets via Kaggle competitions.
• Developed and evaluated models using Python, Scikit-Learn, and XGBoost.
• Gained hands-on experience with feature engineering, hyperparameter tuning, and cross-validation.
Projects Celebrity Face Recognition
|Python, FaceNet, Google Search API April 2025
• Developed a facial recognition model that ranks celebrity lookalikes using FaceNet embeddings.
• Programmatically collected training data for 100 celebrities using the Google Search API.
• Achieved 92%+ accuracy in celebrity face matching using FaceNet embeddings.
• Pro ject link: https://kaggle.com/didartulegen/celebrity-face-recognition
Price Prediction |Kaggle Competition |Python, Scikit-Learn, Pandas, Matplotlib, XGBoost February 2025
• Performed data preprocessing, feature engineering, and exploratory data analysis to understand key price
determinants.
• Used XGBRegressor model to predict backpack prices based on product features such as brand, material, capacity,
and style.
• Pro ject link: https://www.kaggle.com/code/didartulegen/backpack-price-prediction
Model Deployment (OCR + Summarization) |Python, FastAPI, Hugging Face, Google Cloud Run May 2025
• Built a FastAPI service integrating Hugging Face’s TrOCR and DistilBART models.
• Implemented image preprocessing and line segmentation with OpenCV and EasyOCR to improve text extraction
accuracy.
• Containerized the application with Docker and deployed it on Google Cloud Run, configuring CPU, memory,
autoscaling, and health checks.
• Developed REST endpoints and a Jinja2/HTML frontend for file upload, live preview, and JSON-based summary
responses.
• Demo: Live Service Link Technical Skills
Tools & Software
: Python, Java, SQL, Git, Excel
Libraries: Scikit-Learn, Pandas, NumPy, Matplotlib, XGBoost, OpenCV, BeatifulSoup4, Selenium
Actively Learning: GeoPandas, Shapely, Power BI