This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine
What I Built
FINDR AI is an intelligent GPS anomaly detection and investigation system designed to identify suspicious location manipulation in traditional e-commerce parcel delivery systems.
In modern delivery systems, customers and delivery agents depend on GPS coordinates to track parcels and coordinate deliveries. But what happens when the location shown on the screen is not the actual location?
That's where the ghost appears.
FINDR AI detects suspicious GPS behaviours such as:
- GPS Spoofing: Identifies manipulated location coordinates.
- Teleportation Jumps: Detects physically impossible location changes.
- Insidious Drift: Identifies gradual and suspicious location deviations.
- Anomaly Investigation: Investigates suspicious telemetry and assigns risk scores.
- AI Agent Intelligence: Helps analyse suspicious delivery situations and recommend actions.
- Human Approval: Ensures important rescue actions require human authorization.
- Audit Logging: Maintains traceable records of system decisions and events.
I interpreted the Ghost in the Machine theme as a digital ghost hiding inside seemingly normal GPS telemetry — a location that appears real but may not represent reality.
FINDR AI aims to expose these invisible threats before they affect delivery operations.
Demo
Live Dashboard:
https://appropriations-telecharger-candidate-pleased.trycloudflare.com/
Note: This is a temporary Cloudflare Tunnel URL. It is accessible while the local application and tunnel are running.
Video Demo: [Add your demo video link here]
Code
GitHub Repository:
https://github.com/sathvi1234/findr-ai
The project includes the GPS anomaly detection engine, simulated telemetry scenarios, investigation workflows, risk classification, and audit logging.
How I Built It
FINDR AI is built using an open-source-oriented architecture with local backend services and simulated delivery telemetry.
Technology Stack:
- Python: Core anomaly detection and processing.
- Streamlit: Interactive monitoring dashboard.
- FastAPI: Backend API services.
- MongoDB Atlas: Database integration.
- SQLite: Persistent hash-chain audit logging.
- Cloudflare Tunnel: Temporary public access to the locally hosted application.
The system processes simulated delivery-agent GPS telemetry and applies anomaly detection rules to identify suspicious movements.
It combines explainable classification, investigation workflows, risk scoring, human authorization, and audit trails.
The latest regression testing reported 105 passing tests, validating the major detection and workflow components.
Why Does Open Innovation Matter?
Open innovation makes it possible for developers to build, experiment, inspect, and improve intelligent systems without depending entirely on closed platforms.
For FINDR AI, open-source technologies provide:
- Transparency in anomaly detection and decision-making.
- Greater control over sensitive location data.
- Flexibility to experiment with different AI models and agent frameworks.
- Reproducible testing through simulated GPS attacks.
- Opportunities for developers to contribute improvements.
Most importantly, an open approach makes it easier to investigate how a system identifies suspicious behaviour rather than blindly trusting a black-box decision.
My Agent Session
[Add your DevRelay Agent Session link here]
The Spirit of It
Imagine tracking your parcel and seeing that your delivery agent is moving normally on the map.
But suddenly, the location jumps several kilometres away in a few seconds.
The map looks convincing. The coordinates look valid. Everything appears normal until you investigate.
That is the ghost in the machine.
I wanted to build something that looks beyond what a GPS coordinate claims and investigates whether that information can actually be trusted.
FINDR AI is my attempt to make invisible digital manipulation visible, explainable, and actionable.
Prize Categories
[Add only the partner categories whose technologies you actually implemented.]
Acknowledgements
Thanks to DEV Community, Hacktoberfest, and all the challenge partners for creating opportunities to explore open-source AI, experiment with new technologies, and build solutions to real-world problems.
Built with curiosity, experimentation, and a passion for open-source innovation. 🚀
Top comments (0)