Data Science Engineer · 2025-Present

Savelife Ai

SaveLIFE Foundation, New Delhi

1 0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0 1 1 1 1 0 0 1 0 1 1 0 1 0 0 1 1 1 1 1 0 1 0 0 1 1 1 0 1 0 1 1 1 0 0 1 0 0 1 1 1 1 0 0 1

The Challenge

Over 150,000 people die on Indian roads annually. SaveLIFE Foundation monitors 200+ km of national highway corridors, but manually reviewing 500+ hours of dashcam footage was humanly impossible. Audit reports took weeks and stakeholder approvals stalled due to language barriers.

The Approach

Designed a multi-layered AI architecture combining local LLMs (for data sovereignty), RAG pipelines for context-aware violation classification, and MCP servers for orchestrating multi-model inference chains. All dashcam footage stays on Indian servers.

The Solution

AI violation detection pipeline classifying Enforcement, Engineering, and Engagement gaps
Geospatial telemetry engine: EXIF extraction + QGIS + Google My Maps integration
GenAI Translation SaaS for 22+ Indian regional languages (React + Python + MongoDB)
Full-stack dashboard on AWS with 99.9% uptime
Local LLMs ensuring data sovereignty — zero footage leaves Indian infrastructure

The Impact

0+ Hours Processed

0% Detection Precision

0% Manual Effort Reduced

0x Faster Approvals

Tech Stack

Python React MongoDB AWS RAG LLMs QGIS Local LLMs

“When AI can process what humans cannot see at the scale humans cannot match, it stops being a technology story — it becomes a story about saving lives.”

Read the Deep Dive How AI is Saving Lives on Indian Highways →

Next Case Study Arlo Devops →