Research at HOT-G
We are pushing the bits on what we can do with ML on the edge. We are focused on portability, security, pipelines, and conditional decision making.
Containers for edge ML
built on Rust and Webassembly, addressing fragmentation problem.
Mem-safe ML Pipelines
ML pipelines for the Edge including the
authored on Studio guarantee runtime memory safety.
- our core focus to protect the pipeline, models, and the workload on edge with cryptography.
Neural network routing
Conditional processing of complex network setup on the edge. Control blocks for declerative decision making.
Run multiple models in a single pipeline - in series, parallel, or conditional. Great for A/B testing on the Edge.