Is there a way to add custom feature transformation logic before TensorFlow Serving?

Question

According to my research, TFX does not provide support for custom feature transformation within the TFX Pipeline as my custom python functions cannot be serialized inside the gr