Python Developer Needed — Fire Scanner Monitoring

Timeline: 2 weeks

• • ABOUT THE PROJECT

I need a fully functional Python application that monitors live public Broadcastify scanner streams (fire/EMS), transcribes audio in real time, and sends SMS + email alerts when a structure fire, basement flood, or pipe burst is dispatched — then automatically enriches the lead using a Claude AI agent and prepares personalized outreach for my restoration business.

The app should be plug-and-play with any public Broadcastify feed URL, configurable without touching code, and run reliably on Windows as a background service.

• • PHASE 1 — CORE LISTENER (Standalone)
• Audio Capture

— Accept any public Broadcastify stream URL as input

— Support multiple simultaneous streams (multi-county monitoring)

— Chunk audio into 15-second segments continuously using ffmpeg

• Transcription

— Use OpenAI Whisper (local) or Deepgram (streaming) to transcribe each chunk

• Incident Detection & Analysis

— Keyword matching with a fully configurable keywords list (config file, no code changes needed)

— Must detect TWO categories of incidents out of the box:

FIRE:

structure fire, house fire, working fire, flames showing, smoke showing, fully involved, fire alarm, brush fire, confirmed fire

WATER DAMAGE:

water in basement, flooded basement, pipe burst, broken pipe, water main break, sewer backup, water emergency, flooding, standing water

— Anthropic Claude API to confirm the incident type and extract address, incident category (fire vs. water), and responding units

— Must use the official anthropic Python SDK (pip install anthropic)

— Deduplication so the same incident doesn't trigger multiple alerts

• Service Area Filtering

— Only alert if the incident address falls within a configurable list of zip codes or cities

• Alerts

— SMS via Twilio to multiple configurable recipients

— Email alerts with full incident details including incident category (fire or water)

— Configurable quiet hours (e.g. no alerts between 10pm–6am)

• Web Dashboard

— Simple local browser dashboard showing live transcript feed and incident log

— Incident history with address, time, type, category, and which stream it came from

• Logging

— All transcripts and detections saved to a local SQLite database

• • PHASE 2 — CLAUDE AI AGENT (Lead Enrichment)

Once an incident is confirmed, a Claude agent should automatically kick off and complete the following using Claude's tool_use / function calling feature via the Anthropic Python SDK:

— Owner Lookup:

Query county tax assessor records (public data) using the confirmed address to retrieve the property owner name and mailing address

— Contact Enrichment:

Use the Whitepages Pro API to look up a phone number and email for the owner by name + address

— Lead Summary:

Claude compiles a clean lead card (owner name, address, phone, email, incident type, category, time)

— Outreach Draft: Claude drafts a short, empathetic SMS and email tailored to the incident type — different messaging for fire vs.

water damage

— CRM Logging:

Save the full lead card to SQLite and flag it on the dashboard as an enriched lead

— Alert Me:

Send the lead card + drafted outreach via SMS and email for my review

The agent runs automatically after every confirmed incident. If contact info cannot be found, still send owner name and address for manual follow-up.

• • CONFIG FILE SHOULD COVER
• Stream URLs (add/remove feeds easily)
• Keywords per incident category (fire, water)
• Zip codes / cities for service area
• Alert recipients (phone numbers + emails)
• Quiet hours
• API keys including Whitepages Pro (via .env)
• • TECH STACK
• Python 3.10+
• ffmpeg
• OpenAI Whisper or Deepgram
• Anthropic Claude API — claudesonnet46 (incident analysis + agent)
• Claude tool_use / function calling for agent tools
• County tax assessor records (public, per county)
• Whitepages Pro API for contact enrichment
• Twilio
• SQLite
• Flask or FastAPI + basic HTML for dashboard
• Runs on Windows
• • DELIVERABLES
• Clean, wellcommented Python codebase
• for all settings
• .env template for API keys
• README with Windows setup instructions
• 30minute walkthrough call to set it up together
• • IDEAL CANDIDATE
• Experience with realtime audio streaming in Python
• Has built Claude agents using tool_use / function calling with the Anthropic SDK
• Experience with public records lookups and/or contact enrichment APIs
• Has used Whisper, Deepgram, or similar transcription tools
• Familiar with Twilio and basic web frameworks
• Communicates clearly in English

Please include 'scanner' in the first line of your proposal so I know you read this, and share any similar projects you've built — especially Claude agent or data enrichment work.

Contract duration of less than 1 month.

Mandatory skills:

Python, API