GenArchitect

Problem Statement

This project is an AI-powered architecture review assistant designed to streamline and modernize enterprise technical governance. It connects to corporate SSO, applies organizational policies, analyzes domain-specific applications, and retrieves relevant Jira epics and stories to understand technical context. The system allows users to upload architecture documents or Excel-based models for automated evaluation. Using predefined rules and AI reasoning, it detects risks, highlights policy violations, identifies missing controls such as DR, backups, observability, and data protection requirements, and provides actionable recommendations. The final interface presents an Excel-style view with findings, alerts, and improvement suggestions, helping architects make faster and more consistent decisions across the organization.

Solution

This project is built as a modular, web-based architecture review platform powered by AI. The frontend is developed using HTML, CSS, and vanilla JavaScript, structured as a multi-step interface that mirrors the real enterprise architecture workflow. Each screen represents a functional step—SSO, policy filtering, application selection, Jira ingestion, TDD review, and final findings—linked through dynamic navigation logic.The backend is simulated through a combination of mock data and AI-driven logic, where rules, architectural principles, and compliance requirements are processed to evaluate user-submitted files. The system leverages structured prompts and domain-specific policies to generate insights, detect risks, and surface violations such as missing DR, insufficient backups, weak observability, or exposure of sensitive data.To replicate enterprise integrations, the project uses mocked APIs to represent SSO validation, Jira ticket retrieval, and corporate policy feeds. This allowed rapid prototyping without needing real credentials or infrastructure. A lightweight file-ingestion mechanism parses Excel-like inputs and displays them using a custom UI that mimics spreadsheet behavior.Several hacky solutions were used to accelerate development—like dynamic scrolling navigation, flexible data-driven UI rendering, and simulated cloud restrictions (e.g., provider bans, no VPN). These enabled building a realistic architecture-governance tool quickly while maintaining a clean, production-aligned structure.