AI-Driven Automated Code Review and Debugging Using Large Language Models

Software systems are growing in scale and complexity, and ensuring their quality through manual code review and debugging has become increasingly time-consuming, inconsistent, and dependent on the availability of experienced reviewers. This paper presents an AI-Powered Automated Code Review and Code Debugger, referred to as BugFix.AI, which combines static code analysis, machine learning, and large language model (LLM)-based reasoning to detect and correct programming errors with minimal human intervention. The proposed system parses submitted source code, classifies defects such as syntax errors, logical flaws, performance bottlenecks, and non-standard coding practices, and generates context-aware fixes together with human-readable explanations. Suggested corrections are validated through sandboxed execution before being presented to the user, ensuring that no new errors are introduced. The system was implemented using a React.js front end, a Node.js/REST API back end, and GPT-4o for reasoning, and was evaluated on a curated dataset of code samples spanning multiple programming languages. Experimental results show that the system achieves an overall bug-detection F1-score above 74%, reduces manual review effort by a substantial margin, and was rated helpful by a majority of surveyed users. The findings indicate that AI-driven automated review and debugging tools can meaningfully improve developer productivity and code quality while remaining a complement to, rather than a replacement for, human oversight.