Open roles

What we are building

LaniakeaCare is the AI-native clinical software (HIS) used by public hospitals, private clinics and independent health professionals across Colombia. This is not a pilot or a demo: hospitals are treating real patients on it every day.

A single codebase serves two very different realities: independent professionals on a multi-tenant pool, and hospitals on dedicated instances. What you write today runs in production this week.

Why it is hard (and worth it)

The domain is genuinely hard: electronic health invoicing (FEV-RIPS, Resolution 0948 of 2026), integrated care pathways (RIAS 3280), ICD-10/CUPS/CUM catalogues and Colombian data-protection law (Ley 1581, Resolution 1995). A bug here is not a misaligned pixel — it is an invoice rejected by the tax authority or a medical record that fails an audit.

Isolation between hospitals is enforced with PostgreSQL row-level security, not a `where` clause you might forget. AI is not a chatbot bolted on the side: it lives inside the clinical flow (dictation, note structuring, voice-driven orders, PDF reading, specialty-specific agents).

How we work

Small team, no layers: you talk directly to the person who decides the product. We ship often and roll back fast.

We use AI for real, not as a talking point: Claude Code is part of the daily development flow and code lands quickly. So what matters here is not typing more — it is knowing how to DIRECT the tool and prove the result with tests and data, rather than reading lines one by one.

Repo rules are written down and non-negotiable (tenant isolation, audit logging on every medical-record read, never showing invented data on screen). They live in the codebase itself: you read them on day one.