GitHub IntelliSolve: AI-Powered GitHub Ticket Analyzer & Automated Resolver

Abstract: The rapid growth of software development activity on GitHub has resulted in an ever-increasing volume of issue tickets, pull request discussions and support queries that demand timely and accurate resolution. Manual triaging of these tickets places a significant cognitive burden on engineering teams, leading to backlogs, inconsistent labelling and delayed responses. This paper presents the design and implementation of GitHub IntelliSolve, an end-to-end AI-powered system that automates the full lifecycle of GitHub ticket management. The system ingests GitHub Issue and Pull Request events in real time via webhooks, preprocesses and classifies ticket text using fine-tuned BERT and DistilBERT models, performs root cause analysis through GPT-4 and LLaMA-2, generates targeted code suggestions and fixes using Codex and InCoder, and resolves support queries through a Retrieval-Augmented Generation (RAG) pipeline backed by a vector- embeddings knowledge base.

Overview — The proposed system is implemented as a modular, webhook-driven microservice architecture deployed on a Django REST backend with React-based developer dashboard. The pipeline processes incoming tickets through five sequential stages: ingestion and preprocessing, BERT-based classification, GPT-4/LLaMA-2 root cause analysis, Codex/InCoder code generation, and RAG-based contextual Q&A resolution. A continuous monitoring and retraining loop ensures the system remains accurate as codebases and issue patterns evolve.

Result — The expected outcome is a robust, scalable and production-ready intelligent DevOps tool capable of reducing manual ticket resolution effort by over 60%, achieving issue classification accuracy exceeding 92%, and improving mean time-to-resolution for GitHub issues in active open-source and enterprise repositories.

Keywords: GitHub automation, issue triaging, BERT, DistilBERT,GPT-4, LLaMA-2, Codex, RAG, DevOps, NLP, microservices