Portfolio
Academic Activities
RAship in the Dept. of CSE at PSU
- Summary: Since August 2023, I have been a member of the AI4CI core group, where our research focus spans performance and sustainability issues of trend applications, architectures, and various scale distributed systems. My current research interests are listed below.
- UVM Performance in Heterogenous Computing Environments:
- Analyzing performance tradeoffs between page migration and remote direct accesses in multi-GPU systems.
- Mitigating the performance overhead of address translation impact in multi-GPU systems.
- Sustainability in Computing Systems
- Carbon footprint measurements in various scale computing infrastructures.
- Research on the methodological approaches to mitigate carbon footprint on cloud infrastructures.
- LLM Inference Accelerators
- UVM Performance in Heterogenous Computing Environments:
RAship in PARS Research Lab
- Summary: Actively worked on two projects during my RAship at PARS. The former one was about fault tolerance and error propagation analysis of GPGPU applications, while the latter one was about resource-aware compiler design for approximate computing techniques in GPGPU applications. During my RAship there, I gained a comprehensive knowledge of GPU architectures, simulators, applications, and compilers. Our publications and my MSc thesis through these projects are mentioned in the publications section.
- Start/End: 06.15.2020 / 07.10.2023
TAships
- Task: Working actively as a TA in the Computer Engineering Department @IZTECH.
- Start/End: 02.15.2021 / 07.10.2023
- Courses: 1. Computer Architecture (CENG311), 2. Operating Systems (CENG322), 3. Heterogenous Parallel Programming (CENG443), 4. Embedded Systems (CENG424), 5. Theory of Computation (CENG213), 6. Probability and Statistics (CENG222), 7. Human Computer Interaction (CENG318), 8. Information Security (CENG418), 9. Cryptography (CENG471)
CERCIRAS Summer School (2022 - Split, Crotia)
Summary: - I attended the CERCIRAS Training School in 2022. The training school included dense and fast lessons, as listed below, given by many professors from DTU, the University of Padua, RUC, and the University of Szeged.
Time-predictable Multicore Architectures: Architecture introduction and time-predictability; T-CREST multicore processor; reconfigurable interconnectors, and MPI using Argo for NoCs.
Enery Consumption and Optimization Software: Measuring energy consumption through RAPL, and some tools on Python for statistical analysis.
Simulation and Behaviour Analysis of Complex Systems: Simulating IoT-Fog-Cloud systems; modelling, realizing and analyzing fog-assisted blockchain systems; and PriFoB: a Blockchain-based credential management service.
Understanding Mixed-Criticality Systems’ Models and Implementation: Basics of real-time scheduling and response time analysis, Current innovations on mixed-criticality systems, concrete implementation with feather-weight instrumentation.
HiPEAC Student Challenge - The most Efficient Histogram Equalization (2021)
Summary: I attended one of the HiPEAC Student Challenges at Lyon in 2022, where we were expected to implement the fastest and the most energy-efficient histogram equalization algorithm. In this manner, my teammates and I implemented the histogram equalization algorithm on a CPU, GPU, and Xilinx Zynq Ultrascale+, and demonstrated our results on performance and power consumption for those three platforms. We won the competition among other competitors with our results.
Industrial Activities
Job Experiences
Software Engineer at VESTEL Electronics (Manisa, Turkey)
- Task: Develop and maintain the interaction of internet services of TV software.
- Start/End: 05.10.2021 / 10.10.2021
Internships
Sundance Multiprocessor (Chesham, UK) - Long-Term
- Task: Development of Power Profiling for Current Applications through VCS-1 and LynSyn Lite
- Start/End: 08.10.2020 / 02.15.2021
Internship Summary: During my internship at Sundance, I was expected to measure runtime power dissipation and overall energy consumption of contemporary robotic applications that run on Vision, Control and Sensor (VCS) single-board-computers with Xilinx Zynq Ultrascale SoCs. Sundance developed a power measurement tool, LynSyn Lite, measuring both current and supply voltage. To measure the power consumption, I first targeted image processing and deep learning application domains and implemented histogram equalization and MLP algorithms through the Vitis/Vivado environments. I tested my developments in AWS-2 instances provided by Sundance and embedded my application synthesis into the VCS board, which utilizes petalinux as the OS. Afterward, I profiled the power consumption of those applications and reported the results to the company.
Roketsan (Defense Industry - Ankara, Turkey)
- Task: Developing FAST and BRIEF algorithms through Xilinx Ultrascale SoCs
- Date: 2019 Summer Term
Internship Summary: During my internship at Roketsan, I was expected to implement FAST (Features from Accelerated Segment Test) and BRIEF (Binary Robust Independent Elementary Features) algorithms onto Xilinx FPGAs to test how they can be used to guide missile systems. Since it’s a defense industry, I wasn’t allowed to share code developments but I reported the allowed information in the following report.
Dalgakiran Compressor (Compressor Machinery Industry - Istanbul, Turkey)
- Task: Designing and integrating electrical panels, including multiple sensors and PLC, into the compressor machinery.
- Date: 2019 Summer Term
Internship Summary: During my internship at Dalgakiran, I was expected to design electrical boards for various compressor machines where electrical boards are equipped with multiple sensors and a PCB. Basically, PLC controls the machinery actuators through sensors and enables them to operate properly.