Harsh Sharma@harsh_sharma547

Kestrel is a privacy-first, on-device person recognition and access-control system designed for secure physical spaces. A user can be enrolled using only a few reference images, creating a lightweight identity gallery without requiring model retraining for every new person. When someone approaches a protected area, Kestrel captures their image, extracts identity features, matches them against the enrolled gallery, and checks the recognized identity against that area’s access permissions. The AI recognition pipeline runs locally on the Snapdragon X Elite, providing fast, low-latency decisions while keeping sensitive identity data on-device rather than sending it to the cloud. The Arduino UNO Q bridges these AI decisions with the physical environment, handling sensors and access-control components such as LEDs, alarms, and electronic locks. Together, the devices form a complete edge-AI system that can recognize, verify, and physically respond to access attempts in real time.

🌎 Hybrid Hackathon | 💻 Online Build: September 10–16, 2026 | 📍 On-site: September 15–17, 2026 The official hackathon of AI Infra Summit — co-organized by Kisaco Research and lablab.ai. Build production-grade AI applications on the modern AI infrastructure stack, alongside 8,000+ attendees, researchers, founders, and infra leaders. 📍 On-site Venue (September 15–17): Santa Clara Convention Center 5001 Great America Pkwy, Santa Clara, CA 📍 On-site participation is by invitation only. Travel and accommodation expenses will not be covered.