🚀 4 weeks. Here's what we shipped.

I've been building Valiron — an open source AI compliance validation library. Here's the technical breakdown:

What's inside:
→ Binary classification metrics with bootstrap confidence intervals (1000 iterations)
→ Subgroup bias detection across demographic groups — exactly what CDSCO §5.3 demands
→ Calibration analysis (ECE, MCE, reliability curves)
→ Algorithm Change Protocol scaffolding — mandatory for AI/ML SaMD updates
→ CDSCO MDSW report generation — section-by-section mapped to actual regulatory requirements

Numbers:
✅ 106 tests passing
✅ 94% code coverage
✅ MIT Licensed
✅ Zero patient data leaves your system

The problem: Clinical AI companies spend 3–6 months + ₹30L on compliance documentation. We automated it.

One command:

If you're building AI for healthcare, finance, or HR — and need CDSCO, FDA, or EU AI Act compliance — check it out.

GitHub: github.com/valiron-ai/valiron
PyPI: pypi.org/project/valiron

#MachineLearning #AI #OpenSource #BuildingInPublic #HealthcareAI #Startup