Publications

Summary. A software article describing bdd2pw, an open-source tool that scaffolds runnable Playwright TypeScript tests from Gherkin .feature files. It detects existing Page Objects, inspects the live DOM through the Microsoft Playwright MCP, and emits Page Object Models and spec files ready for execution, with an optional governed LLM fallback. Published as @vijaypjavvadi/bdd2pw on npm and archived on Zenodo.

V. P. Javvadi, R. S. Ganapathy, M. S. Sri, S. Pochincharla, T. V. Akula, et al.

Summary. A practitioner-oriented decision catalogue for recognising, classifying, and mitigating flaky tests in continuous-integration pipelines, organised by root-cause family (timing, order dependence, shared state, environment, and infrastructure) with recommended detection signals and remediation strategies for quality-engineering teams.

Abstract. Systematic leave-one-repository-out validation reveals an AUC–F1 asymmetry under cross-repository conditions and identifies defect-rate mismatch as the primary qualitative predictor of transfer degradation. Cross-repository AUC ranges 0.817–0.913 (mean 0.867); cross F1 ranges 0.525–0.707 (mean 0.631).

Abstract. Top-k coverage analysis at k ∈ {10, 20, 30, 40, 50}% with risk-vs-random lift factors. The top 10% of files ranked by predicted risk capture 43.82% of defects (4.37× lift over random). The deployed model is integrated as a FastAPI microservice triggered from GitHub Actions, with explicit advisory-vs-gating rollout criteria.

Abstract. A framework that recovers broken UI locators at runtime using DOM similarity analysis and tree-ensemble ranking over eight feature families (id-stability, role/aria, structural neighbourhood, text proximity, depth, attribute Jaccard, sibling index, and XPath edit distance). A mutation-based evaluation pipeline simulates realistic UI evolution across 2,400 mutation events in seven refactoring classes; hybrid heuristic + ML ranking yields a deterministic top-1 candidate with calibrated confidence to gate auto-heal versus flag-for-review.

Summary. An executable mutation study asking whether LLM-generated tests for community-maintained open-source services are better grounded in natural-language requirements or in source code, measured against an independent human-written baseline. Extends the earlier RAITG line of work (see Preprints below).

Summary. Evaluates what a cross-project defect predictor is actually worth to a practitioner on 1,323 real, reproducible defects from Defects4J and BugsInPy — framing the decision as trust the predictor, augment human judgement with it, or replace it, with calibration and cost-aware analysis.

Summary. A deterministic timeline engine for detecting employment-experience fraud at scale, paired with an adversarial robustness study against AI-generated fakes. The TruthHire reproducibility bundle provides the reference implementation, evaluation harness, and a public-corpus assessment (CC-BY-4.0).

Summary. A threshold private-set-intersection architecture that lets organisations report and cross-check fraud signals without revealing their underlying records, an outcome-symmetric incentive design that aligns reporting behaviour, and an empirical parameterisation on real fraud data.

Summary. Measures how often LLM-generated test data reproduces published fixtures verbatim rather than synthesising novel values, across five models and eight public schemas — with implications for test-data provenance, licensing, and the validity of LLM-generated fixtures in CI.

Abstract. The RAITG framework converts 312 requirements across three domains (banking, healthcare, e-commerce) into executable tests using an LLM plus deterministic rule-based verification: 68% effort reduction (184 h → 58.9 h), requirement coverage 71.2% → 94.1%, first-pass verification rate 96.3%. Symbolic mutation indicators serve as a domain-agnostic adequacy proxy. A reframed successor study is under review at Wiley STVR.

Abstract. An empirical comparison of four post-execution defect-attribution methods on 6,000 synthetic failure events, fusing dynamic test-failure signals (stack traces, failure clustering, suspect-set ranking) with repository priors from the defect-prediction model to produce actionable triage advice rather than ranked file lists. A successor study on 1,323 real defects is under review at PeerJ Computer Science.

Public preprint of the manuscript currently under review at Elsevier Computers & Security (listed above).

Abstract. Uniform empirical comparison of six classifiers (LR, DT, RF, GB, XGB, MLP) on a 296,457-instance dataset assembled from five mature open-source systems (Elasticsearch, Spring Boot, Hadoop, Kafka, Express). Stratified 5-fold cross-validation with paired-t and Wilcoxon significance testing. Random Forest is best (AUC 0.8998); XGBoost is the closest competitor (AUC 0.8955). Includes multi-model Gini feature importance with explicit caveats on high-cardinality bias.

Summary. A mixed-methods study of how teams adopt, migrate between, and augment browser-automation frameworks (Selenium, Playwright, Cypress, Puppeteer, WebdriverIO, Nightwatch, TestCafe) with AI agents — combining a practitioner survey, GitHub repository mining, and package-registry metrics across npm, PyPI, Maven, and NuGet.

Summary. A controlled factorial experiment (fault location × guard arm) using the agentic-qa runtime on three open-source web applications, with an execution-based weakening verdict, to test whether LLM test agents silently weaken test suites and whether enforced quarantine outperforms instruction-only guards. Pre-registration planned.

Summary. A software article for a deterministic, AST- and rule-based migration toolkit that converts Selenium test suites (Java TestNG/JUnit/Cucumber) to Playwright TypeScript scaffolds. Published as @vijaypjavvadi/sel2pw on npm and archived on Zenodo.

Summary. Short software articles for four open-source packages published under the @vijaypjavvadi npm scope: pw-emit (shared emitter for Playwright artefacts from a source-agnostic IR), usagecontract (usage-aware consumer-driven contract testing for HTTP APIs), synthdata (LLM-driven synthetic test data with referential integrity), and agentic-qa (an auditable agentic QA runtime with a governance gate and replayable ledger).