Utku U.
Deep Learning Specialist
Experience
Combining Neural Fields with Hypernetworks
- Developed a meta-learning approach with a teammate to merge multiple neural fields into a single scene representation using a hypernetwork.
- Implemented and evaluated the method on 2D (MNIST) and 3D (ShapeNet) data, showing faster inference compared to overfitting-based baselines.
Data Analytics for Yacht Chartering
- Collaborated in a team of four to build predictive models and a dashboard for yacht chartering as a part of an interdisciplinary project at TUM.
- Conducted exploratory data analysis (EDA), feature engineering, and clustering (HDBSCAN) to define sailing areas.
- Developed demand forecasting, pricing, and segmentation models to support yacht owners and charter companies in strategic decision-making.
Career Advisory in VR
- Created a VR game that runs on browsers to aid students in their career decisions with realistic and immersive simulations as part of a practical course on serious games in extended reality at TUM.
Big Data Intern
HAVELSAN
- Learned to use open-source software that is built to process and stream large-scale data, like Apache Spark and Apache Kafka.
Intern
Parton Big Data Analytics and Consulting
- Designed and built web applications that serve different purposes with HTML, CSS, JavaScript, and Python.
- Created a console application for cloud services with various features, such as AWS EC2 instance creation and a fully working SSHv2 terminal for EC2 instances.
- Developed a dashboard application for visualizing the results of an artificial intelligence system that predicts quality scores of production in real time.
- Trained, evaluated, and compared different deep learning models like LSTMs and Transformers for time series forecasting problems with PyTorch.
Undergraduate Research Member
ITU Vision Lab
- Finished online courses and lectures for computer vision with deep learning.
- Completed the material for the lecture “Deep Learning for Computer Vision” from Michigan University.
- Completed my bachelor’s thesis on Novel Class Discovery in a continual learning setting.
- Read and studied both fundamental and recent publications on the topics of continual learning and self-supervised learning.
- Designed and tested various deep learning models with PyTorch for improving the performance of state-of-the-art methods by integrating new ideas from self-supervised learning.
Summary
I am a deep learning specialist with a strong background in neural network research and data analytics. My work spans meta-learning, computer vision, and time series forecasting, where I have built and evaluated models that enhance inference and decision-making. I have a solid grasp of both practical and theoretical aspects of deep learning, continually integrating the latest research into my projects.
I bring hands-on experience in training deep models using real-world data from diverse fields such as scene representation, yacht chartering, and VR simulations. I excel in using frameworks like PyTorch and big data tools, and I focus on delivering practical, scalable, and efficient solutions.
Languages
Education
Technical University of Munich
MSc · Informatics · Munich, Germany · 1.5/1.0
Istanbul Technical University
BSc · Computer Engineering · İstanbul, Turkey · 3.82/4.0
Certifications & licenses
Deep Learning Specialization
Coursera
Flutter & Dart - The Complete Guide [2025 Edition]
Udemy
Machine Learning
Coursera
Similar Freelancers
Discover other experts with similar qualifications and experience
- English
- Deutsch
2025 © FRATCH.IO GmbH. All rights reserved.