Résumé

HackMol 6.0, team of four

CivicSentinel.

Emergency reporting that turns a web form, text or voice note into a validated police report (FIR), triaged by AI and anchored on a blockchain.

Code on GitHubLive demo

Filing a police report is slow, hard to follow and easy to alter after the fact. CivicSentinel lets citizens report the way they already communicate, has AI turn that into a structured report, and records a hash of the evidence on Ethereum so a quiet edit can't go unnoticed.

  • Ways to report

    3

    web form, SMS or a voice note.

  • Evidence trail

    On-chain

    hashes on Ethereum, files on IPFS.

  • My commits

    21 of 44

    top contributor on the team repo.

How it works

From report to record.

  1. Report

    A citizen files through the web app, by SMS through Twilio, or with a voice note, after signing in with a one-time password.

  2. Structure

    A LangChainJS pipeline with GPT-4 turns free-form audio or text into a validated FIR record, asks for missing details, and scores severity.

  3. Anchor

    Evidence is pinned to IPFS through Pinata and its hash is written to an Ethereum smart contract, so any later change to the record is detectable.

  4. Act

    Officers work from a priority-ordered dashboard, and background jobs escalate cases that miss their deadline.

  5. Monitor

    API latency, uptime and health are tracked in Prometheus and Grafana, with logs shipped to Loki.

Tech stack

What it's built with.

Frontend

React, TypeScript
The citizen and officer apps
Tailwind CSS, Radix UI
Interface components
TanStack Query
Server state and caching
React Hook Form, Zod
Validated report forms
Leaflet
Maps for incident locations

Backend

Node.js, Express
The REST API
MongoDB, Mongoose
Reports, users and case history
Redis
Caching and background task queues
JWT, bcrypt
Sign-in and sessions
Helmet, express-rate-limit
Hardening the public API
Twilio
SMS reporting and OTP sign-in
Multer, Cloudinary
Evidence uploads
node-cron
Deadline checks and escalation

AI and blockchain

LangChainJS, GPT-4
Report structuring and severity triage
Solidity, Hardhat, Ethers.js
The evidence contract and its tooling
Pinata (IPFS)
Content-addressed evidence storage

Operations

Prometheus, Grafana
Latency, uptime and health dashboards
Winston, Loki
Structured logs
Docker, PM2, AWS EC2, Vercel
Running and hosting it

My part

What I built.

  • Built the intake pipeline (Twilio, GPT-4, LangChainJS) that converts unstructured audio and text into validated FIR records with automated severity triage, and anchored evidence hashes on Ethereum smart contracts and IPFS for a tamper-proof audit trail.
  • Added OTP-based citizen authentication and background task queues, and monitored API latency, uptime and health with Prometheus and Grafana.

Next project

Zenith

Hiring for backend, forward-deployed or GPU work? Let’s talk.

saxena.shikhar1808@gmail.com

GitHubLinkedInLeetCodeRésumé