PAIMANA Infrastructure Intelligence

PAIMANA is a public infrastructure monitoring and predictive risk analytics platform.It combines a React frontend,FastAPI backend,machine-learning risk scoring, early-warning alerts,role-based access.

Comment

Inspiration

Addressing Overruns: Public infrastructure projects in India frequently face severe delay and cost overruns across multiple ministries and states.

Proactive Monitoring: Inspired to move away from static administrative records toward a unified, predictive monitoring tool for administrators, inspectors, and citizens.

What it does

Portfolio Analytics: Interactive dashboards showing high-risk projects, expenditures, sector views, and early warnings.

Predictive Risk Engine: Computes delay probabilities, cost overrun estimates, and recommended mitigation actions using ML models.

LogicCore AI Assistant: Role-aware chatbot powered by Gemini (with Groq and local database failovers) to query project statuses and alerts securely.

Role-Based Access: Multi-tier access for Super Admins, Field Inspector Officers, and Citizens/Public Viewers.

How we built it

Frontend: Built with React 19, Vite, Recharts, and Lucide React.

Backend: Asynchronous Python API powered by FastAPI, Pydantic, and Uvicorn.

Database & ML: SQLite persistence (paimana.db) with XGBoost and scikit-learn risk models (risk_model_v2.joblib).

Challenges we ran into

Database Persistence: Handling SQLite on ephemeral cloud filesystems like Render by configuring volume mounting (MOSPI_DATABASE_PATH).AI Provider Availability: Designing a 3-tier fallback chain (Gemini $\rightarrow$ Groq $\rightarrow$ Local DB fallback) to ensure continuous chat support.Role Scoping: Enforcing context isolation in LLM queries to keep data private across different user tiers.

Accomplishments that we're proud of

Real-time automated risk calculations triggered upon updating or adding projects.

Fail-safe AI assistant execution capable of citing project dossiers directly.

Granular RBAC that smoothly separates administrative tasks from field inspection workflows.

What we learned

End-to-end integration of machine learning pipelines into REST API workflows.

Prompt context security to prevent prompt-injection data leaks across user roles.

Robust dynamic audit logging for tracking state changes and image-backed inspector comments.

What's next for PAIMANA Infrastructure Intelligence

PostgreSQL Migration: Upgrading from local SQLite to managed PostgreSQL for scalable production.

GIS & Mapping Integration: Adding map overlays for regional risk clustering and spatial project tracking.

IoT & Remote Sensing: Incorporating satellite imagery and field sensors for automated progress verification.

Built With

Share this project:

Updates

Submission history