Evangale Jooste
@EvangaleKTV
Independent AI safety researcher building SEVERANT, a formally verified, hardware-constrained AI architecture where unsafe behaviour is physically impossible by design.
https://github.com/EvangaleKTV/SEVERANT/tree/main$0total balance
$0charity balance
$0cash balance
$0 in pending offers
About Me
I am a sole independent researcher based in Cape Town, South Africa, building SEVERANT from the ground up with no institutional affiliation and no external funding to date.
SEVERANT is a 7-layer AI architecture where safety is a hardware property, not a training outcome. Its ethical constraint layer is formally verified in Lean 4, physically write-locked into silicon, and active throughout every training step of every other layer.
A working prototype is operational. The formal verification suite is complete. The knowledge base is under active construction. Everything built by one person.
Full architecture and documentation at the GitHub link above.