Join our open-source movement as an intern. Earn a verifiable joining letter, an ID card, and valuable
real-world experience.
How it Works
01
Discover Opportunities
Explore the repositories listed below. You can either submit your own ideas or work on existing issues.
02
Reach Out
Contact us at admin@mnmworks.xyz with your details, ideas, or the repository you want to work on.
03
Join as an Intern
Get accepted as an intern, earn a valid joining letter and verifiable ID card, and start building.
Active Repositories
// open source
StreetIQ
Active
An AI-powered road intelligence & navigation platform that turns every driver into
a road surveyor. Real-time hazard detection using a custom-trained YOLOv8 model running entirely
in-browser via TensorFlow.js.
ReactTensorFlow.jsYOLOv8Node.jsSupabaseWebSockets
// android app
Kavach
Active
An on-device cybersecurity and fraud prevention Android app engineered for senior citizens. Automatically intercepts scam SMS, inspects WhatsApp messages, analyzes unknown APKs before installation, and alerts trusted family contacts offline.
The AI-Powered Road Intelligence & Navigation Platform
StreetIQ is a next-generation navigation platform that tells you what is
actually on the road. It provides real-time turn-by-turn navigation overlaid with live
crowdsourced hazard data potholes, waterlogging, cracks, and debris detected entirely on-device using a
custom-trained YOLOv8 model running in the browser at 15fps via TensorFlow.js. Video feeds are never sent
to a server. Only GPS coordinates and hazard metadata are transmitted, preserving absolute privacy.
Custom-Trained YOLOv8 Model
Hand-trained on thousands of real Indian road images to detect cracks, potholes, waterlogging, and
debris across varied lighting and traffic conditions.
On-Device AI — Zero Video Uploads
All inference runs locally in your browser. Nothing leaves your device except GPS coordinates when
a hazard is confirmed.
Real-Time Hazard Broadcasting
The moment a hazard is detected, WebSockets instantly broadcast it to every other driver in the
area via the Supabase Realtime backend.
Full Navigation Stack
Smooth turn-by-turn navigation powered by OSRM with custom traffic multipliers and Photon geocoding
— no third-party map API required.
Make your changes in the forked repo — bug fixes, new features, improvements, anything meaningful.
Raise a Pull Request with proper documentation describing what you changed and why.
Head to the Submit section of your MnM Dashboard, paste your fork link, and write a
description of your contribution.
Wait for review, approval, and feedback from the MnM team.
If it is taking longer than expected, mail us at admin@mnmworks.xyz for faster validation and review.
// open source project
Kavach
Digital Safety & Scam Protection for Senior Citizens · वरिष्ठ नागरिकों के लिए डिजिटल सुरक्षा
Kavach ("shield" in Hindi) is a privacy-first Android app safeguarding senior citizens from digital financial fraud — OTP scams, fake KYC / account-blocked SMS, extortion threats, lottery baits, and malicious APKs distributed via WhatsApp. Operating 100% offline with zero telemetry, Kavach scores threats locally, reads out warnings aloud in plain English or Hindi using on-device text-to-speech, and instantly alerts a trusted family member when high-risk activity is detected.
Automatic SMS Interception
On-device BroadcastReceiver scans incoming SMS in real time, detecting phishing links, banking extortion, urgency patterns, and OTP solicitations with zero cloud transmission.
WhatsApp & Clipboard Scanner
One-tap verification for suspicious forwarded WhatsApp and Telegram messages with bilingual risk assessment (LOW, SUSPICIOUS, HIGH) and audio readout in English and Hindi.
Pre-Install APK Analysis
Deep static manifest analysis inspecting dangerous permissions (READ_SMS, accessibility services, device admin) to flag and block banking trojans before installation.
Trusted Contact & Emergency SOS
Dispatches privacy-safe SMS alerts to a designated family member on high-risk detections, with one-tap dialing for National Cybercrime Helpline 1930 and Chakshu reporting.
Active
Active
