Efficient edge AI,
from model to silicon.
Kalanthyx develops accelerator IP and the compiler that targets it, designed together as one system for on-device inference.
01
The stack
IN
Model
Standard models from today's major ML frameworks.
KALANTHYX
Compiler
Our own toolchain, from model to hardware.
OUT
Accelerator IP
Synthesizable RTL, configurable to your power, performance and area targets.
02
Contact
We're working with a small number of early partners.
If you're building silicon for the edge, developing on-device AI applications, or investing in the space, we'd like to hear from you.