Show HN: I built a Raspberry Pi webcam to train my dog (using Claude) (github.com)

A product manager built "YogiCam," a Raspberry Pi webcam pipeline (Pi + camera module, Python with PiCamera2, Flask frontend, HTML/CSS) with help from Claude to monitor and time her dog's separation-anxiety training. After swapping to PiCamera2 from an older libcamera suggestion, she got livestreaming working locally, added a built-in stopwatch to timestamp barks for a trainer, and used ngrok to create a public URL so monitoring works over cellular. The system cost about $127, includes a mini tripod and case for stability, and reduced Yogi’s barking from a 3-second panic to 30+ minutes of calm within six weeks alongside training and medication. The project highlights how LLMs speed hobbyist engineering and rapid prototyping for real-world behavioral applications. Key technical takeaways: Flask serves the live video, PiCamera2 is the reliable camera library, ngrok solves NAT/Wi-Fi limits, and adding a USB mic + audio-streaming and automatic bark detection (dB-threshold logging) are straightforward next steps. The author also suggests simple automation (physical button triggering a bash script to start the server, ngrok, and SMS the link). YogiCam is a compact example of edge devices + LLM-assisted development delivering practical, low-cost solutions for monitoring and telemetry in animal behavior and beyond.