A recent Snapshot of New Generation Coding Practices with AI Models

Title: A recent Snapshot of New Generation Coding Practices with AI Models

Speaker: Atay Ozgovde

Time/Date: 11:00 / 13.03.2026

Location: VYKM-2 Seminar Room, 5th Floor, Faculty of Engineering, South Campus

Abstract: Large language models have moved beyond code completion into agentic, multi-step software engineering tasks, yet empirical understanding of their real-world impact remains fragmented. This talk synthesizes recent findings from practitioner studies, software engineering benchmarks, and controlled experiments to characterize the current state of AI-assisted coding. We survey emerging patterns across the stack: from prompt engineering and context management strategies, to evaluation gaps between benchmark performance and production utility, to the systemic risks introduced by agentic pipelines including cascading errors, over-trust, and the gradual erosion of deep debugging expertise. We close with open research questions and directions the field has yet to address.

Short Bio: Dr. Atay Özgövde is currently a full-time faculty member in the Department of Computer Engineering at Boğaziçi University.  His main areas of expertise are Edge and Cloud Computing, Applied AI for Communication Networks, Game Programming, and Generative AI-Based Next-Generation Software Development Methods. His latest research focuses on designing efficient LLM/SLM inference components on edge-cloud continuum for coding related workloads.