Understanding and Detecting Platform-Specific Violations in Android Auto Apps
Moshood A. Fakorede, Umar Farooq

🏛️ IEEE/ACM AST 2026 — Oral Presentation

Built AutoComply, a static analysis framework using a Car-Control Flow Graph (CCFG) to detect compliance violations in Android Auto apps. Detected 27 violations across 31 apps — 13× more than Android Lint.

MobileDev-Bench: A Benchmark for Issue Resolution in Mobile Application Development
Moshood A. Fakorede, Krishna Upadhyay, A.B. Siddique, Umar Farooq

⏳ In Submission

A benchmark of 407 real-world issue-resolution tasks across 19 production mobile apps (Android, React Native, Flutter). Frontier LLMs achieve only 3.39–5.21% resolution rates, exposing a critical gap in mobile SE capability.

How Far Can LLMs Go in Generating Android Lint Checks from Natural Language?
Moshood A. Fakorede, A.B. Siddique, M. Sridharan, Umar Farooq

⏳ In Submission

LintBench, a 113-check benchmark evaluating LLMs on generating Android Lint checks from natural language. Across four LLMs, API retrieval and execution-guided repair raise success from 30%–65% to 57%–85%. Yet most compiled checks that fail miss required issues, making correct issue-detection logic the central challenge.

Understanding Bugs in Quantum Simulators: An Empirical Study
Krishna Upadhyay, Moshood Fakorede, Umar Farooq

⏳ In Submission

An empirical study of 394 confirmed bugs across 12 open-source quantum simulators. We find that most failures surface post-deployment, logical bugs often produce silently incorrect outputs, and many critical issues trace back to classical infrastructure rather than quantum logic itself.