Community Discussions
Building a self-hosted health data platform — looking for testers
Hey everyone,
I've been building something I call a "personal health operating system" — a self-hosted platform that pulls together data from multiple wearables and apps into one dashboard, with an AI assistant that actually understands your data in context.
What it does
The core idea: no more jumping between Whoop, Withings, Apple Health, and five other apps. Everything lands in one place, and an AI agent can reason across all of it.
Multi-source aggregation — Whoop, Withings, Yazio, Hevy (native integrations) + Elite HRV and ActiBreathe (manual input). Apple Health is supported via Health Auto Export — a third-party iOS app that costs €8/year or €30 lifetime. I have no affiliation with them, just the best solution I found for getting Apple Health data out.
AI Chat with tool-calling — queries your health data, training history, nutrition, and uploaded documents in real time (supports Ollama locally or OpenRouter for cloud models)
Research Agent — automatically fetches and scores papers from PubMed, bioRxiv, and medRxiv relevant to longevity, biohacking, and your specific supplement stack
Document Intelligence — upload blood work PDFs, doctor's letters, DEXA scans — AI extracts summaries and makes them searchable in chat
Health Context Profile — auto-generated health snapshot injected into every conversation so answers are actually personalized
Muscle Heatmap — SVG body map colored by training volume across muscle groups
Configurable dashboard with charts, mobile layout, and full DE/EN localization
Tech stack
Python/FastAPI backend, vanilla JS frontend (no framework overhead), PostgreSQL. Runs on any always-on Linux box — I use a Mele Quieter 3 mini PC alongside Home Assistant. Cloudflare Tunnel handles OAuth callbacks without opening ports.
Why I'm posting
Full transparency: this is my first real software project. I'm not a developer by background, and I've been building this with AI assistance — which means it works well for my setup, but there are almost certainly rough edges, blind spots, and things I haven't thought to test. I'm planning to open-source this in the coming weeks, but before that I'd love a small group of technically minded people to stress-test it and call out what's broken, misconfigured, or just badly designed.
What you'd need
A Linux server or Raspberry Pi (always-on)
PostgreSQL
At least one supported data source
~30 minutes for initial setup
If you're into quantified self or biohacking and comfortable with a bit of self-hosting, drop a comment or DM me. I'll help you through setup personally and actively incorporate feedback before the public release.
Hey everyone,
I've been building something I call a "personal health operating system" — a self-hosted platform that pulls together data from multiple wearables and apps into one dashboard, with an AI assistant that actually understands your data in context.
What it does
The core idea: no more jumping between Whoop, Withings, Apple Health, and five other apps. Everything lands in one place, and an AI agent can reason across all of it.
Multi-source aggregation — Whoop, Withings, Yazio, Hevy (native integrations) + Elite HRV and ActiBreathe (manual input). Apple Health is supported via Health Auto Export — a third-party iOS app that costs €8/year or €30 lifetime. I have no affiliation with them, just the best solution I found for getting Apple Health data out.
AI Chat with tool-calling — queries your health data, training history, nutrition, and uploaded documents in real time (supports Ollama locally or OpenRouter for cloud models)
Research Agent — automatically fetches and scores papers from PubMed, bioRxiv, and medRxiv relevant to longevity, biohacking, and your specific supplement stack
Document Intelligence — upload blood work PDFs, doctor's letters, DEXA scans — AI extracts summaries and makes them searchable in chat
Health Context Profile — auto-generated health snapshot injected into every conversation so answers are actually personalized
Muscle Heatmap — SVG body map colored by training volume across muscle groups
Configurable dashboard with charts, mobile layout, and full DE/EN localization
Tech stack
Python/FastAPI backend, vanilla JS frontend (no framework overhead), PostgreSQL. Runs on any always-on Linux box — I use a Mele Quieter 3 mini PC alongside Home Assistant. Cloudflare Tunnel handles OAuth callbacks without opening ports.
Why I'm posting
Full transparency: this is my first real software project. I'm not a developer by background, and I've been building this with AI assistance — which means it works well for my setup, but there are almost certainly rough edges, blind spots, and things I haven't thought to test. I'm planning to open-source this in the coming weeks, but before that I'd love a small group of technically minded people to stress-test it and call out what's broken, misconfigured, or just badly designed.
What you'd need
A Linux server or Raspberry Pi (always-on)
PostgreSQL
At least one supported data source
~30 minutes for initial setup
If you're into quantified self or biohacking and comfortable with a bit of self-hosting, drop a comment or DM me. I'll help you through setup personally and actively incorporate feedback before the public release.
Please sign in to post a reply.
