🛰️ GeoSense: Spatiotemporal Telemetry Anomaly Detection Engine

🌐 Live System & Interactive HUD

Lead Geospatial Forensic Expert: Nikhil Krishnan (NK)

Interactive 3D Control Center: nikhilkrishnan.dev

Mobile App Simulator: Truth_detector

Professional Profile: LinkedIn — Nikhil Krishnan

🧠 Executive Summary

GeoSense is an autonomous geospatial diagnostic and anomaly interception framework designed to detect GNSS / GPS coordinate manipulation and hardware-level signal spoofing.

Typical security systems blindly trust location payloads. GeoSense bypasses this vulnerability by executing low-level spatiotemporal mathematical checks against real-world physical limits, protecting tracking data in real-time.

🚀 The Architecture: 2-Machine Production Sync

This project runs continuously on a dedicated, physical dual-node local infrastructure integrated with multi-cloud resources:

                 [ Azure Cloud Database ] 
                            |
                            | (Fetch 76,135 Telemetry Rows)
                            v
                 +-------------------------+
                 |   GEEKOM CLIENT NODE    | (Windows/WSL Client)
                 |   - pipeline.py Engine  |
                 +-------------------------+
                            |
                            | (Secure LAN Route via Port 5432)
                            v
                 +-------------------------+
                 |   UBUNTU SERVER NODE    | (Local IP: 192.168.10.4)
                 |   - Docker Containers   |
                 |   - PostGIS Database    | (project_data / geonode_data)
                 |   - Cockpit Management  | (Port 9090 Console)
                 +-------------------------+

🛰️ The Data Pipeline Flow:

Edge Ingestion: Telemetry data is simulated via Android hardware nodes (Termux) and posted securely to a serverless gateway.

Google Cloud Run API: Serves as a secure API Proxy that processes requests and routes AI queries to Gemini-2.5-Flash (Vertex AI Platform) without exposing client-side API keys.

Queueing & Lakehouse: Streams live packets through Azure Event Hubs and directly into Microsoft Fabric Eventstream Delta Tables.

Local Database Pour: The python sync pipeline pulls the raw 76,135 spatiotemporal rows and bulk-injects them directly into my local Dockerized PostgreSQL/PostGIS database!

🕵️‍♂️ Case #250-KM: Impossible Physics Isolation

During a real-world spatiotemporal audit, GeoSense isolated a critical GPS spoofing attempt on an active vehicle:

The Anomaly: The vehicle coordinate points made an instantaneous 250 kilometer jump from an Abu Dhabi residential street directly into the offshore Ruwais oil sector.

The Math: The temporal gap between the pings was under 0.8 seconds, indicating an impossible physical velocity of 1,474 km/h.

The Isolation: By calculating the Euclidean distance delta and passing it through a Moving Average Buffer, the system automatically identified the signal as "COMPROMISED" and quarantined the payload.

⚙️ Core Stack & Technologies

Languages: Python, SQL, JavaScript (ES6+), Shell Scripting

Databases: PostgreSQL / PostGIS, SQL Server, Azure SQL

Cloud & DevOps: Google Cloud Run, Vertex AI Studio, Azure Event Hubs, Microsoft Fabric, Docker Compose, GitHub Actions, Linux / Ubuntu Admin

Frontend Graphics: Three.js (3D WebGL), Tailwind CSS, Chart.js

🏆 Verified Credentials & Badges

🛡️ Vertex AI Prompt Design (Google Cloud)

🛡️ Gemini Enterprise Agent Ready (Google Cloud)

🛡️ ChatGPT Prompt Engineering (DeepLearning.AI)

🛡️ Diploma in Warehouse Management (Alison, Distinction)

🛡️ 100% English Proficiency Score (EF Standardized Assessment)