How would you structure a JNI bridge to pass user text input to a C++ based LLM inference engine and return the generated response?
