I Put My Son in a Swamp

a day ago

The author modeled his newborn son as a typed resource in a system called 'swamp,' which includes a sync method for pulling the last seven days of data, reports summarizing sleep and feeding, and idempotent delete functionality.
He developed multiple interfaces to track the baby, including a mobile app, a Pebble watchapp (for which he lacks the hardware), a conversational server, Home Assistant for reminders, and a swamp extension, all connecting to the same Baby Buddy REST API.
The tracking system serves as a shared source of truth for two exhausted parents, facilitating handoffs during shifts by providing accurate, real-time data on feeding, sleep, diapers, and medication without needing to wake each other.
Key engineering lessons emerged: ensuring data completeness (e.g., using a 'truncated' flag to indicate missing records) and designing for human error (e.g., soft-deleting timers to avoid data loss from race conditions or mistakes).
The author open-sourced the extension on the swamp registry, noting that the instincts driving this project also apply to his work at Shrug PW, where stakes are higher but sleep is better.

Hasty Briefsbeta