AI-powered IT analytics platform built in a 24-hour hackathon to predict and flag operational risks in software project management.
Delivery risk in software projects usually surfaces as a missed deadline — long after the signals that predicted it appeared in daily logs, bug trends, and task churn. Project managers are reacting to outcomes instead of leading indicators.
Treat project telemetry as a time series and score it continuously. Rather than another reporting dashboard, the system watches team activity for anomalies and surfaces a ranked risk view answering one question: which projects need attention this week?
Full-stack monorepo. Next.js frontend for the analytics dashboards, FastAPI service exposing scoring and anomaly endpoints, PostgreSQL for project and activity history, and ML risk models running behind the API. Containerised with Docker; designed for Jira and Kaizen ingestion.
A working end-to-end prototype delivered inside a 24-hour sprint — risk scoring, anomaly detection, cost-overrun forecasting, and automated reporting, with role-based access already in place.