İsmail Can Dikmen Electrical & Electronics Engineering · İstinye Üniversitesi, Istanbul

Two weeks of hands-on research with three high school students, Zeynep Elveren, Zeynep Bahat and Yusuf Kaymakçı, in the Neuromorphic Algorithms track of İstinye University's Genç Bilim İnsanları (Young Scientists) summer research internship, organized end to end by İstinye TTO. The outcome was three papers on spiking neural networks: two submitted to the 10th International Artificial Intelligence and Data Processing Symposium (IDAP 2026), and the third to the 30th IEEE High Performance Extreme Computing Conference (HPEC 2026).

The titles do not read like high school work, and that was exactly the point. In two weeks the students learned scientific methodology by applying it directly: setting the question up right, and going back to the method when a result did not come out as expected. A result that looks good is not automatically a correct one; they saw this firsthand. The credit is theirs.

Designed and delivered ZeroTech Academy, TEMSA's EV and next-generation vehicle technologies program, together with Teoman Karadağ. About six months went into building the curriculum department by department, followed by six months of delivery across sites in three countries: the Adana plant and R&D center, Istanbul, Germany, and France. Content spanned battery systems, electric powertrains, charging infrastructure, and next-generation vehicle architectures, reaching nine functional departments from R&D through Sales (WE and EMEA), Operations, and support units.

Effectiveness was measured, not assumed. A quasi-experimental pre-test/post-test design was applied to 315 employees with a validated instrument (expert-reviewed content validity, Cronbach's α > 0.70). Mean scores rose from 59.5 to 70.2 on a 100-point scale, an 18% gain that was significant on a paired-sample t-test (t = 2.35, p = 0.019). Knowledge gain correlated with operational performance (r = 0.45, p < 0.01), and a department-level ANOVA (F = 1.85, p = 0.055) showed R&D and Sales responding most. Analysis ran in SPSS under ethics-committee approval, and the results were published in JEBI (2025).

Presented NEUROSEC at the 20th International Symposium on Applied Reconfigurable Computing (ARC 2024) in Aveiro, Portugal. The work brings neuromorphic computing to audio security on FPGA, reaching a 94% threat-detection rate and holding up against FGSM and PGD adversarial attacks at a 5.39 dB SNR. The full paper appears in the ARC 2024 proceedings (Springer LNCS, pp. 134–147).

Presented at the 27th International Conference on Computing in High Energy & Nuclear Physics (CHEP 2024) at Jagiellonian University, Kraków. The talk covered neuromorphic and hybrid neural-network models for particle physics, mapping spiking architectures onto FPGA hardware through HLS4ML for high-throughput, low-power inference. CHEP is the flagship venue for computing in high-energy and nuclear physics.

Invited speaker at the 4th Automotive Summer Camp (Otomotiv Yaz Kampı), the online program run by the Automotive Manufacturers Association of Türkiye (OSD) and supported by OİB. Delivered the session on the past-to-present arc of electric and hydrogen vehicles together with Teoman Karadağ, to university students exploring careers in the automotive industry. The camp draws the sector's leading manufacturers and reaches a large national student audience each year.