Mehmet Rıfat Özkurt

• Conducted research on Retrieval-Augmented Generation performance in VLMs through representation learning methods. • Implemented state-of-the-art RAG pipelines with industrial datasets and multimodal LLMs. • Created benchmark datasets for RAG evaluation.

• Tutored the programming exercises for Embedded Systems and Security, a course of 300 to 400 students per semester, helping students write and debug machine-level software exploits in the lab. • Maintained the exercise content and its build system, including the Makefiles, and aided in course organization. • Diagnosed, repaired and tested the faulty development boards.

• Designed a VCO and a cascaded filter to operate at the UHF band. • Implemented and optimized the design with respect to parameters such as phase noise, output power, bandwidth, frequency resolution and pass-band flatness. • Assisted in debugging old RF designs.

• Investigated FPGA-based novel technologies on clock synchronisation and syntonization. • Developed FPGA-based serial communication modules.

• Implemented a simple Pulse-Doppler Radar with signal processing tools on Zynq-7000 SoC. • Optimized the data transfer through designing and implementing a DMA between PS and PL fabric.

Thesis: “Extensions to Deep Learning Methods for Full Waveform Inversion in Non-Destructive Testing”