CV

Master Thesis Researcher — Multiview Panoptic Operating Room Segmentation

Informatics 16, Technical University of Munich

• Investigated 3D feature lifting using NeRFs and voxel networks for panoptic segmentation.
• Explored vision foundation models (SAM, DINO, CLIP) for multimodal understanding.
• Adapted VGGT’s aggregator for spatial understanding, achieving +5% Panoptic Quality (PQ) improvement with the video model and +7% PQ improvement with the multi-view model on MM-OR and 4D-OR datasets.

Intern / Working Student — Deep Learning for Communications

Huawei Research Center

• Built an autoencoder FFN (PyTorch) with a model-based decoder for signal covariance estimation, outperforming baseline by ~+7 dB SNR.
• Built an automated synthetic data generation, model training and evaluation pipeline.
• Trained a transformer model with unique learnable position embeddings for beamspace phase error estimation achieving similar performance as the PNPC algorithm in significantly lower time.

Research Intern — Semantic Mapping with Quest 3 AR/VR

Chair of Cognitive Systems, Technical University of Munich

• Developed a Unity (C#) app for 6-DoF tracking and scene mesh extraction on Quest 3.
• Integrated ROS# for real-time wireless transfer to master computer for ray tracing from headset pose in the scene mesh with <10 ms latency.
• Deployed Mask2Former (Detectron2) on the master computer for real-time panoptic segmentation, achieving mesh labeling at 15 Hz throughput using NVIDIA RTX3090.

Bachelor Thesis Researcher — Shared Control for Robotic Rehabilitation

Chair of Information-oriented Control, Technical University of Munich

• Simulated a 2-DoF robot arm’s dynamics in MATLAB and designed PD + gravity compensation-based impedance controllers for position tracking.
• Implemented an assist-as-needed strategy adapting controller gains online based on human performance.

Working Student — C/C++ Developer

AED Vantage

• Developed C++ modules for the IPBasis ECU of the new BMW vehicle generation.

Part-Time Electronics Engineer

TUfast Student Racing Team

• Designed schematics and layout for centralized A/D boards and microcontroller programming in C++.
• Won 3rd place at Formula Student Germany (FSG) 2022.

M.Sc.

Technical University of Munich (TUM)

Electrical & Computer Engineering

• Advanced Non-Linear Control
• Data-Driven Decision Making
• Estimation Theory
• Signal & Image Processing
• Master Thesis: Multiview Panoptic Operating Room Segmentation (Grade: 1.0)

B.Sc.

Technical University of Munich (TUM)

Electrical & Computer Engineering

• Circuits and Systems
• Control Systems
• Computer Systems
• Signal Theory
• Stochastics
• Analog and Digital Circuits
• Linear Algebra
• Bachelor Thesis: Shared Control Based on Both Tracking Performance and Human Torque for Robotic Rehabilitation (MATLAB/Simulink, Grade: 1.7)

Abitur

Deutsche Schule Istanbul

German High School

PanORama: Multiview Consistent Panoptic Segmentation in Operating Rooms

arXiv

Multiview consistent panoptic segmentation framework for operating room scene understanding.

SAM 2026 (Under Review)

Transformer-based deep unfolding approach for phase noise compensation in beamspace DOA estimation.