Two courses at five universities in the 2026–27 autumn term, under the Ministry of Industry and Technology's “Industry on Campus” programme.
Why most transformation programmes stall, and how governance, prototypes and phased delivery get them moving again. Where AI agents and generative models create value in enterprise processes.
How an e-commerce product is run from strategy to release: roadmap, prioritisation, experimentation, marketplace and platform decisions. Taught through cases from real programmes.
ORCID 0009-0007-7884-2872
1st International Edessa Scientific Studies Congress (Harran University / ISARC), Şanlıurfa, Türkiye · Proceedings pp. 365–374 · 6 June 2026
A framework that reuses in-store infrastructure (security cameras, POS, door counters, energy and fitting-room sensors, time attendance) to monitor store operations in real time, with inference at the edge to protect privacy. Over a six-month pilot in twelve stores, peak checkout waits fell by about 25%, fitting-room abandonment by about 20% and energy use outside staffed hours by about 15%, while coverage of underserved store zones rose by about 30% (observational figures). This is the academic study of the store camera analytics project that went live in all of the group's stores.
10th International Mediterranean Scientific Research and Innovation Congress (ISARC), Antalya, Türkiye · Proceedings pp. 193–204 · 12 April 2026
Recasts loyalty programmes as engagement systems built on dynamic tier-based segmentation, incentive-driven behavioural design and engagement-triggered interactions, drawing on a real omnichannel implementation. After launch, purchase frequency, basket value, cross-channel engagement and upward tier movement all trended up (directional results).
12th International Hasankeyf Scientific Research and Innovation Congress (ISARC), Batman, Türkiye · Proceedings pp. 533–540 · 31 January 2026
A production-grade multi-model generative AI architecture (diffusion image generation, transformer NLP, MLOps) for product images, SEO content, advantage text and review-driven size recommendation. Evaluated on about 121,000 SKUs and over one million reviews, it cut content production time by about 70%, manual intervention by up to 60% and external content cost by 40–50%. The size recommender reached an F1 score of 0.79 and showed potential to reduce size-related returns by 5–8%.
8th International Selçuk Scientific Research Congress (ISARC), Konya, Türkiye · Proceedings pp. 1023–1029 · 30 November 2025
An automated decision-support system for e-commerce order fulfilment at FLO (about 25,000 orders a day). Built on Go microservices that communicate over RabbitMQ, it assigns warehouses with the Analytic Hierarchy Process and carriers by proportional distribution on performance, cost and return rate. After deployment, warehouse assignment accuracy reached 97% and manual carrier intervention fell from 6.7% to 0.07%. Decisions took 10 ms instead of 15–20 seconds, deliveries were 14% faster and logistics cost fell by 6.3%.
Guest lectures and conference talks on e-commerce product management, digitalisation governance and applied AI.
