A division of Johnson & Johnson, one of the world’s largest and most diversified healthcare corporations. The medical devices group designs and manufactures innovative surgical, orthopedic, and interventional solutions used by healthcare professionals in over 60 countries.
Project Overview
Developed a distributed analytics platform for Johnson & Johnson Medical to measure real time visitor engagement during the EADV 2025 conference. EventScope combines edge computing, computer vision, and data visualization to provide actionable insights into booth performance, dwell times, and attendee behavior, all while maintaining complete privacy compliance.
The Challenge
Event organizers needed accurate, real-time metrics to understand how healthcare professionals interacted with content and staff across multiple booth locations.
Key goals included:
- Operating in venues with limited network connectivity
- Ensuring privacy compliance (no PII or cloud uploads)
- Supporting 20+ concurrent installations
- Delivering sub-2s dashboard latency
- Automated deployment with minimal technical setup
Technical Architecture
Edge Layer
- Custom IoT hardware with integrated camera and compute module
- On device ML inference for emotion, gesture, and interaction detection
- Offline first local store with batch sync and exponential backoff
Backend Layer
- Rails 8 Solid Stack (Solid Queue, Cache, Cable)
- PostgreSQL partitioned event schema with deduplication
- Real time WebSocket updates using Hotwire Turbo Streams
Visualization Layer
- Live dashboard displaying visitor analytics, engagement rate, dwell time, sentiment, and interaction counts
- Built with Hotwire, Tailwind CSS, and ViewComponent
- Optimized for 4K displays and touch interfaces
Key Features
- Real time visitor tracking across multiple booths
- On device anonymization and privacy first architecture
- Zero touch provisioning and network auto discovery
- Schema driven validation for reliable data ingestion
- Automated 30 day retention and cleanup system
Results
- 10,000+ events/day processed with sub 100ms latency
- 95%+ uptime under real world event conditions
- Delivered actionable analytics to event staff in real time
Technologies Used
- Backend: Rails 8, Ruby 3.4, PostgreSQL, Solid Queue/Cache/Cable
- Edge: Python 3.11, FastAPI, TensorFlow Lite, MediaPipe, SQLite
- Frontend: Hotwire, Tailwind CSS, ViewComponent
- Infrastructure: Custom IoT hardware, WireGuard VPN, Ansible Automation
Takeaway
EventScope showcases how privacy first, edge driven systems can achieve real-time analytics at scale, even in challenging, connectivity-limited environments.
It proved that high performance ML inference, low latency data aggregation, and live visualization can coexist entirely within an on-premise architecture built for reliability and compliance.