Duygu Yalçınkaya is an ISTQB® Advanced Level certified Senior Software Test Engineer with more than 10 years of experience in software testing across e-commerce, legal tracking systems, internet security, network monitoring systems, and educational technology domains.
She has hands-on experience in functional testing, API testing, web services testing, test planning, test case management, regression testing, user acceptance testing, and non-functional testing areas such as performance, usability, and compatibility.
Throughout her career, Duygu has worked closely with development, product, and cross-functional teams to improve software quality throughout the development lifecycle. Her current interests focus on how AI-assisted development changes the role of QA, especially around non-functional risks such as performance, security, reliability, and production readiness.
Speech: When AI Writes the Code: Why Non-Functional Testing Becomes More Critical Than Ever
How AI testing agents can support performance, security, and reliability in modern QA workflows
AI-assisted development is rapidly changing how software teams design, implement, and deliver products. Developers can now generate backend services, APIs, database queries, integration logic, and even test cases faster than ever before. While this creates a significant productivity advantage, it also changes the quality risks that QA teams need to focus on.
As code is produced faster, functional correctness alone is no longer enough. An API may return the expected response, a user flow may work as designed, and automated tests may pass successfully. However, the system may still perform poorly under load, expose sensitive data, fail during timeouts, create scalability issues, or become difficult to monitor and troubleshoot in production.
This session explores why non-functional testing becomes more critical in the age of AI-generated and AI-assisted code, especially for backend-heavy teams. The talk will focus on practical risk areas such as inefficient database queries, missing authorization checks, weak input validation, lack of rate limiting, poor timeout handling, fragile integrations, and insufficient logging or observability.
The session will also connect these risks to the conference theme by discussing how AI testing agents can support modern QA workflows. Rather than replacing QA expertise, testing agents can help teams identify non-functional risks earlier, generate performance test scenarios, support security review checklists, analyze logs and metrics, evaluate reliability concerns, and assist with production-readiness reviews.
The talk will also address an often-overlooked question: how do we test the testing agents themselves? If an AI agent suggests performance scenarios, identifies security risks, or summarizes release readiness, QA teams need a way to evaluate the quality, traceability, consistency, and limitations of those outputs.
Participants will leave with a practical perspective on how AI-assisted development changes the software quality risk map, why non-functional testing deserves more attention, and how AI testing agents can support performance, security, reliability, and production-grade quality in modern QA workflows.
