Ring Cerberus: Autonomous Perimeter Lockdown

• 

  Commercial Landing Page with Enterprise Security Architecture & SaaS Tiers

• 

  Tactical Command Console with Real-time MediaPipe Edge Telemetry & Noise Floor

• 

  Threat Scenario 1: Package Poacher Detected with AI Verbal Intercom Warning

• 

  Threat Scenario 2: Armed Forced Entry with Red Alert & 98.4 dB Acoustic Shield

• 

  Threat Scenario 3: Authorized Courier RFC 6238 TOTP Airlock Verification

Inspiration

What it does

How we built it

Inspiration

Traditional smart doorbells are passive observers: they notify homeowners after a package theft or forced entry has already transpired. Homeowners are forced to manually open an app, wait for a stream to buffer, and verbally intervene. We engineered Ring Cerberus to bridge the gap between passive alerting and active cyber-physical defense—transforming Ring devices into a self-governing, multi-agent security sentinel that neutralizes threats in real time.

What it Does

• Edge-First Motion Gating: Runs client-side MediaPipe Wasm computer vision to filter out 84%+ benign environmental frames before hitting cloud reasoning, drastically cutting latency and token overhead.
• Autonomous Interrogation Swarm: When suspicious activity crosses confidence thresholds, an AWS Bedrock (Claude 3.5 Sonnet) agent initiates real-time verbal de-escalation over the intercom.
• Rolling Ambient Acoustic Sentinel: Employs dynamic noise-floor normalizers to identify glass breaks, crowbar strikes, and forced entries while suppressing background urban noise.
• Zero-Trust Delivery Airlock: Integrates an offline-resilient TOTP vault (RFC 6238) enabling couriers to deposit parcels securely into smart drop-boxes without full perimeter access.
• Enterprise Tactical HUD & Dispatch: Provides an interactive cyber defense console with manual tactical overrides, acoustic strobes, and automated Twilio/SNS emergency dispatch alerts.

How We Built It

• Cloud Reasoning: AWS Bedrock (Claude 3.5 Sonnet / Haiku) orchestrating multi-agent threat triage.
• Edge Intelligence: MediaPipe Wasm for localized browser/device computer vision gating.
• Backend Engine: FastAPI (Python 3.11) with WebSockets delivering real-time telemetry streams.
• Frontend Console: Next.js 14, Tailwind CSS, Lucide Icons, and Web Audio API.
• Production Infrastructure: Fully deployed on Vercel with automated GitHub CI/CD workflows.

Challenges We Ran Into

• Cloud Inference Cost vs. Latency: Continuously streaming high-res video to Bedrock was cost-prohibitive. We engineered an edge-first gating pipeline that lowered cloud cost by 84.6% and saved an estimated $142.30/mo.
• Acoustic False Positives: Differentiating wind gusts from physical impacts required implementing a rolling noise-floor normalizer.
• Zero-Trust Offline Resilience: Ensured standard TOTP algorithms run entirely client-side so deliveries can be processed even if local internet drops.

Accomplishments That We're Proud Of

• Built an end-to-end cyber-physical security platform within 37+ files (~3,000 lines of robust code).
• Achieved zero TypeScript/hydration warnings in production Next.js compilation.
• Live production deployment accessible globally with sub-second dashboard telemetry.

What's Next for Ring Cerberus

• Native Ring Chime Pro integration with local acoustic waveforming.
• Automated 911 / Central Station dispatch through UL-certified monitoring API connectors.
• Matter / Thread protocol support for offline smart-lock coordination. ## Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Ring Cerberus: Autonomous Perimeter Lockdown

Built With

• aws-bedrock
• claude-3-5-sonnet
• fastapi
• mediapipe
• nextjs
• python
• react
• tailwindcss
• typescript
• vercel
• webrtc