In this post, we will introduce Spark-TFRecord, a new solution to enable support for native TensorFlow data format in Spark.