Show HN: SynapseKit – Async-native Python framework for LLM pipelines and agents
github.com
2 threads
Async is definitely the right approach for this, but how are you handling the event loop blocking if a local model takes way too long to generate a heavy sequence?