Spark framework is way different and runs on many nodes. If it is Spark Streaming (Talend Real Time Big Data), you can publish the data to kafka / JMS and pick it up in normal talend job.. If it is Spark batch, you can not do this. You may have to pass via a HDFS file.