Xubru.AI
AI-Powered Business Intelligence Platform
Xubru.AI is a business intelligence platform that enables teams to upload data and instantly generate dashboards, KPI templates, AI-powered summaries, and deep analytics insights. As an intern, I worked directly on the production codebase — identifying and resolving high-impact bugs in the authentication system, billing flows, and conducting a comprehensive security audit.
System Design
Xubru.AI uses a decoupled architecture with a Next.js frontend served via Nginx reverse proxy, and a Python FastAPI backend handling data processing, AI summarisation, and REST API endpoints. Services are orchestrated with Docker Compose.
01Frontend
Next.js + React application handling dashboards, KPI template selection, file uploads, and real-time chart rendering. Communicates with the FastAPI backend via REST.
02Backend
Python FastAPI service responsible for data ingestion, AI-powered analysis, user authentication (OAuth), billing/coupon logic, and serving analytics results to the frontend.
03Infrastructure
Docker Compose orchestrates the frontend, backend, and database containers. Nginx acts as a reverse proxy routing traffic and handling SSL termination in production.
04Security
Conducted a full security audit covering OAuth token handling, session management, API endpoint authorization, and input validation. Identified and patched several critical vulnerabilities.
What it does
Problems solved
Diagnosed and fixed a broken OAuth flow where tokens were not being invalidated correctly on logout, leaving sessions exposed
Resolved a coupon code creation bug in the billing system that allowed invalid coupons to be applied silently
Conducted a full security audit — identified insecure direct object references on several API endpoints and patched them
Hardened API routes against unauthenticated access by auditing middleware chains across the FastAPI service