Saranya Alagarsamy

Saranya's doctoral research tackled one of the longstanding challenges in software testing: automating the generation and maintenance of unit test cases. Writing effective unit tests is labour-intensive and requires deep understanding of the code under test — its structure, semantics, and expected behaviour in different contexts.

The thesis developed context-aware deep learning approaches that go beyond syntactic code analysis, incorporating semantic and contextual signals to generate test cases that are not only compilable but also meaningful — targeting relevant behaviours and edge cases. A critical and often overlooked dimension of the work is test repair: as source code evolves, existing test cases can become stale or broken. Saranya's methods address both the generation and the ongoing repair of unit tests in living codebases.

This work is at the frontier of AI-assisted software testing, contributing techniques that reduce the burden on developers while maintaining — and potentially improving — the quality of the test suite. The research was conducted at Monash University under the supervision of A/Prof. Tantithamthavorn and A/Prof. Aleti, with Dr Arora as associate supervisor.