Open to opportunities

melis günşeber

@melisgneber

AI/ML engineer focused on safe LLM agents and production computer vision—building efficient edge models with PyTorch and YOLO.

Germany

Message

What I'm looking for

I’m seeking a full-time AI engineering role where I can build and deploy production-ready ML systems—combining efficient computer vision/edge inference with research-driven focus on LLM safety and trustworthy AI.

I’m an AI/ML engineer with 3+ years of progressive experience, from deploying production computer vision systems to building real-world robotics and smart-retail solutions. I’ve developed production-grade object detection for real-time fraud detection and designed a stand-alone smart vending machine, including image processing, object detection, and sensor fusion deployed on Nvidia Jetson.

At DFKI, I worked as an AI working student on COMET to decentralise Smart Living data—helping teams publish accessible datasets via a metadata catalogue and improving stakeholder clarity through interactive prototypes. I’m also completing an M.Sc. in Artificial Intelligence at TU Darmstadt, with a master’s thesis focused on LLM agent safety via sparse autoencoders, combining safety-driven research with engineering that’s ready for deployment.

Experience

Work history, roles, and key accomplishments

AI Working Student

DFKI

Apr 2025 - Mar 2026 (11 months)

Contributed to the BMBF-funded COMET project to decentralise Smart Living data, enabling individuals to control personal data while allowing access to high-quality datasets for AI innovation. Built a public landing page with a metadata catalogue, developed a click prototype to visualise historical data requests, and created technical documentation using Antora.

Python PyTorch Antora Metadata Technical Writing Landing Pages literature review Cross Functional Collaboration

AI Specialist (Robotics IoT)

MİGROS TİCARET A.Ş.

Aug 2023 - Sep 2023 (1 month)

Developed a production-grade YOLOv8 (PyTorch) object detection model to identify irregularities at supermarket cash registers for real-time fraud detection. Benchmarked deep learning frameworks to optimise inference speed and accuracy for edge deployment.

YOLOv8 PyTorch Object Detection Computer Vision Benchmarking Real Time Systems

Junior AI Specialist (Robotics IoT)

MİGROS TİCARET A.Ş.

Jul 2022 - Jul 2023 (1 year)

Designed and developed a stand-alone smart vending machine from the ground up, implementing image processing, object detection, and sensor fusion to detect and track products reliably. Integrated the machine with a mobile app for secure transactions and reduced product misidentification via iterative model fine-tuning and hardware-software co-optimisation on Nvidia Jetson.

Object Detection Image Processing NVIDIA Jetson Model Fine Tuning Sensor Fusion

AI Intern (Robotics IoT)

MİGROS TİCARET A.Ş.

Nov 2021 - Jun 2022 (7 months)

Built an object recognition ML model for retail product detection to serve as the foundation for the smart vending machine system developed in the subsequent role. Supported reliable product detection through iterative model development and preparation for downstream integration.

Object Detection Machine Learning Model Development data preprocessing Model Training Model Evaluation Python

ITU Lab Intern (Thesis Project)

Istanbul Technical University

Aug 2021 - Jun 2022 (10 months)

Developed an algorithm using textile-based sensors and motion capture to quantify a dancer’s foot movement performance, enabling objective skill assessment. Applied time-series and sensor-driven analysis to convert motion signals into measurable performance scores.

Textile Sensors Motion Capture Time Series Analysis Algorithm Development Signal Processing Machine Learning Objective Assessment

Education

Degrees, certifications, and relevant coursework

Technical University of Darmstadt

Master of Science, Artificial Intelligence and Machine Learning

2023 -

Activities and societies: Zuse School Eliza Scholar (Oct 2024 – Mar 2026).

Currently completing an M.Sc. in Artificial Intelligence and Machine Learning at TU Darmstadt. Master’s thesis focuses on safeguarding LLM agents via latent space manipulation with sparse autoencoders.