Case study
Interactive AI Hologram Installation
Led backend and AI integration for an interactive hologram installation presented at ITEX 2024. The installation received the event’s Best Booth award.
Role
Technical & AI Team Lead
Technology
- Python
- AI/ML
- FastAPI
- WebSocket
- Docker
Context
An interactive hologram installation needed to turn a camera feed into a responsive projected experience at a public technology event.
Problem
The pipeline had to connect perception, application control, and projection without making the presentation depend on a fragile collection of unrelated scripts.
Constraints
The system ran in an event environment with changing lighting, a bounded compute budget, and a visible user experience that had to fail gracefully.
My role and ownership
I led the technical and AI work, shaped the service boundary, and connected the perception pipeline to the projection control path with a small delivery team.
Architecture
Camera → pose estimation → API/WebSocket control → projection. The perception step produced bounded application events; the projection layer remained responsible for presentation and fallback behavior.
Key decisions and trade-offs
Keeping the model-facing work outside the projection renderer made the visible experience easier to control. A simpler event contract was preferred over exposing model internals to the presentation layer.
Failure modes and safeguards
The system could fall back to a stable presentation state when camera input, inference, or transport was unavailable. The public record does not establish a measured latency benchmark, so no latency number is claimed here.
Observed outcome
Outcomes
Presented at ITEX 2024; the installation received the event’s Best Booth award.
Evidence boundary
The event photograph documents delivery and team context. It is not presented as a product screenshot or a benchmark report.