Write tech design doc for the proposed architecture changes to ComfyStream / ai-runner and proposed inference routing mechanism
with handoff to infra team to get feedback
https://www.notion.so/livepeer/Multithread-inference-1ba0a348568780c7ac67e90dde712b23?pvs=4