Shram Sanchar is a mobile-first web prototype built for the hackathon under the track Migrant Worker Tracking in eShram. It addresses the challenge of keeping migrant workers’ official employment and location records up to date as they relocate across India for work. Unlike traditional solutions that depend on continuous, battery-draining background tracking and consistent high-speed internet, Shram Sanchar implements an opt-in, privacy-centric check-in model that requires no invasive surveillance or complex manual forms.

At its core, the platform prioritizes accessibility and resilience through voice-first and offline-first capabilities. Workers can simply record a spoken update in their regional language, which is parsed by AI into structured fields such as destination, trade, and journey status. Designed to function in low-connectivity conditions, all check-ins are saved locally on the device using IndexedDB and automatically synchronized once network access is restored. Privacy is built in by default: location services remain disabled until explicitly toggled on by the worker, reading coordinates only once per confirmed check-in and rounding them to roughly 1 km in alignment with the DPDP Act, 2023.

Beyond basic check-ins, the application includes a dedicated Complaint Desk, an interactive administrative portal, and a multilingual digital assistant. Workers can register workplace grievances—such as wage theft or safety hazards—with automated AI triage that assigns urgency tiers and suggests emergency helplines when critical risks are identified. For labour officers, the administrative dashboard provides live metrics, complaint lifecycle tracking, interactive maps of verified worker movements, and CSV export tools. The integrated assistant, Shram Sahayak, offers instant conversational guidance on labor rights, schemes, and helplines.

The solution is implemented using a lightweight, performant stack featuring React 18, HTML5, CSS3, and JavaScript, paired with Leaflet.js, OpenStreetMap, and the Nominatim API for spatial visualization. It leverages native browser interfaces including the Web Speech API, Geolocation API, MediaDevices, and IndexedDB, backed by Node.js/Express and Google Gemini models for voice parsing and grievance classification, and is deployed live via Vercel.