Project
- Discipline
- Software
- Timeline
- June 2026Ran as a service on a Mac hub until August 2026
- Role
- Designer & builderSolo
- Category
- Health technologySelf-hosted infrastructure
- Status
- v1, on demo dataThe ring is not paired yet
Tools
- Python
- Flask
- SQLite
- NumPy
- SciPy
- Chart.js
- launchd
A self-hosted scoring engine for a $30 smart ring
LinkRing computes WHOOP-style recovery, strain and sleep scores for a $30 smart ring, on hardware I control. It runs end to end on 30 days of seeded demo data; the ring is not paired yet, so there are no real nights in it.
Recovery is measured against my own 14-day rolling baseline, not a population average. Recovery (0–100) maps the night's resting heart rate — the 5th percentile of the overnight window — against that baseline as a logistic z-score. Strain (0–21) weights the day's heart-rate minutes across five effort zones; a day at 14 or above is classified hard. Sleep duration and quality are estimated from the overnight heart-rate dip.
Until August 2026 it ran as two launchd services on a Mac hub — a WSGI server for the dashboard, and a 15-minute cycle that recomputed scores, wrote the last 14 days of scores to my Obsidian vault, and rotated database backups. The cycle's ring sync step was skipped every time, because no ring was paired.
The dashboard
Each day gets an end-of-day report with a plain-language verdict — "well recovered, green light to push" or "run down — favour rest" — and a one-tap export to my Obsidian vault. Metrics the hardware can't measure yet show no number; the card says what's missing and why.



Key decisions
- Personal rolling baseline over absolute thresholds — the scoring engine (~200 lines of NumPy/SciPy) runs eight self-checks on every scoring run.
- Service setup: WSGI server, launchd auto-restart, rotating logs, daily database backups with 14-generation retention.
- Reused the ring's existing open-source client instead of writing custom Bluetooth code.
- Obsidian vault integration: until August 2026, the last 14 days of scores synced into one note in my knowledge base every 15 minutes.
- HRV math (RMSSD) and heart rate from a raw PPG wave are written and self-tested, but not wired in until real-time pulse capture exists to feed them. v1 runs end to end on seeded demo data; the ring is not paired yet.
The whole system is about 625 lines of Python across five files, two SQLite databases, and zero cloud dependencies.
v1, on seeded demo data — the ring is not paired yet