B.S. Electrical & Electronics Engineering · Hacettepe University, 2026
RTL-to-GDSII · RISC-V Architecture · High-Speed FPGA Design
Hacettepe University
B.S. in Electrical & Electronics Engineering
Sep 2022 – Jun 2026 · Ankara, Turkey
3.04
GPA / 4.00
FPGA Design Candidate Engineer (Part-time) · Ankara, Turkey
AR-GE Electronics Design Candidate Engineer · Ankara, Turkey
Electronic System Design Intern · Ankara, Turkey
Led a 4-person team to 4th place among 94 teams at Teknofest 2026 with a multi-platform RISC-V MCU: the same RTL is prototyped on Gowin & Xilinx FPGAs through automated TCL flows and hardened to GDSII via OpenLane/LibreLane on the SKY130 PDK. Includes a lightweight custom AI accelerator; verified with a UVM environment and Spike-ISS co-simulation.
Designed a high-performance, multi-mode 32-bit ALU achieving 1.66 GHz in Turbo mode. Complete RTL-to-GDSII flow using Cadence Genus & Innovus on the GPDK45 (45 nm) PDK. Timing closure and area optimization via custom constraints; functional verification with Cadence Xcelium.
Co-developed a multicycle RV32IM RISC-V processor in Verilog under a 38-state FSM, running at 100 MHz on a Gowin GW2A FPGA and capable of running the original DOOM from compiled C at ~1.2 FPS. Features a DMA bootloader, 4 KB L1 I/D caches (miss penalty cut from ~20 cycles to ~2), HDMI output, PS/2 & UART input, and a hardware crash-dump debug mechanism.
Graduation project — real-time detection system pairing a RISC-V core with a fully custom RTL AI accelerator running the TinyDet model architecture on 128×128 images at 13 FPS. A ping-pong buffer strategy overlaps data transfer with compute. The hand-written RTL was chosen after evaluating Vitis HLS, MATLAB HDL Coder, & hls4ml; models trained with MATLAB Deep Learning Toolbox & Google Colab.
Designed custom PCBs for UAV payload release, power distribution, and electronic cut-off systems within a 10-person team. Panelized for JLCPCB fabrication using Altium Designer. Team achieved 1st place in Turkey and 5th globally.
Designed and fabricated a solid-state Tesla coil with an ESP32-controlled full-bridge driver at 550 kHz. Precision MOSFET gate driving and active feedback control for reliable high-voltage air breakdown.
Custom FPGA-based GPU performing real-time Gaussian blur using parallel computing and custom hardware accelerators.
Pushing a 55 nm Gowin FPGA toward 700 MHz and an Intel/Altera Agilex 3 toward 1 GHz through aggressive custom timing constraints and clock-tree tuning, to validate one overclocking methodology across process nodes and vendors.