SENTRI — Smart Entry with Behavioral Intelligence

A security SaaS prototype for gated communities that learns suspicious visitor behavior patterns and helps guards, residents, and admins act on it in real time.

Overview

Manual visitor registers are low-signal and slow to search, and give operators no memory of repeat behavior. SENTRI replaces that with face-capture-driven logging, an explainable trust-score engine, and anomaly detection — surfaced across three purpose-built interfaces for guards, residents, and admins.

Live demo: https://sentri-app-tau.vercel.app

Features

• Trust Score Engine (0–100) — deterministic score from visit frequency, host diversity, late-night entry timing, repeat denials, and external risk signals; every score comes with a factor-level breakdown for auditability
• Decision bands — SAFE / CAUTION / WATCHLIST / FLAGGED
• Detection stack — Isolation Forest anomaly signals + a deterministic rule engine, combined into a confidence-aware anomaly feed
• Three-role ecosystem
  • Guard — React console for pass scanning and gate-side decisions
  • Resident — mobile-first one-tap approve/deny flow
  • Admin — anomaly monitoring, device visibility, user/access policy, surveillance views
• Role-gated navigation & RBAC via lib/accessControl.ts

Architecture

flowchart LR
  A[ESP32-CAM Gate Unit] --> B[Guard App / Web Console]
  B --> C[Trust Score Engine]
  C --> D[Decision Band SAFE/CAUTION/WATCHLIST/FLAGGED]
  D --> E[Resident Notification]
  D --> F[Admin Anomaly Feed]
  F --> G[Blacklist and Policy Action]

Loading

Tech Stack

Layer	Technology
Framework	Next.js 16, React 19, TypeScript
State	Zustand
Data fetching	TanStack React Query, Axios
Motion/UX	Framer Motion
Charts	Recharts
Auth (demo mode)	Role-based local session model

Demo Walkthrough

Demo credentials (local/demo-mode auth):

Role	Email	Password
Admin	admin@sentri.ai	Admin@123
Guard	guard@sentri.ai	Guard@123
Resident	resident@sentri.ai	Resident@123

1. Open the landing page and read the problem framing + trust gauge.
2. Go to /auth, log in as admin.
3. Open /admin and run Simulate to inject a suspicious visitor.
4. Visit /admin/anomalies to see risk classification.
5. Switch to /guard to verify the pass-scanning flow.
6. Switch to the resident role to see the one-tap approval model.

Getting Started

Prerequisites

• Node.js 18+

Installation & local development

git clone https://github.com/https-shubhamsahu/sentri.git
cd sentri
npm install
npm run dev

Open http://localhost:3000.

Production build

npm run build
npm run start

Project Structure

src/
├── app/
│   ├── page.tsx                 # Landing page
│   ├── auth/                    # Role login/signup
│   ├── admin/                   # anomalies, devices, surveillance, users, visitors
│   ├── guard/                   # Guard operations + QR verification
│   └── resident/                # Resident workflows (visitors, settings)
├── components/
│   ├── ui/
│   └── next/
├── lib/
│   ├── trustEngine.ts           # Scoring logic
│   └── accessControl.ts         # RBAC matrix
└── store/                       # Zustand app state

hardware/                         # ESP32-CAM + operator-console counterpart (see below)

Hardware Companion

SENTRI's edge unit is an ESP32-CAM based capture module (~₹600 estimated hardware cost). The camera firmware, control-board firmware, and Flask/OpenCV operator console live in hardware/ — see hardware/README.md for setup and wiring.

Compliance & Security Posture

Designed around CERT-In IoT security principles, with role-based access control and explainable scoring to avoid opaque decisioning.

Contributing

Fork the repository, create a feature branch, and open a pull request.

License

No license file is currently present in this repository.