The support for RDD is not exposed in the v10.x of Spark Connector Drivers. you still see a dataframe inferred using https://www.mongodb.com/docs/spark-connector/current/batch-mode/batch-read/#schema-inference
You have two options - (a) Explicitly specify the schema instead of relying on the auto-inference, (b) Increase the number of documents that are sampled for auto-inferring the schema (however this will lead to proportional increase in initial time latency) .
sampleSizeThe number of documents to sample from the collection when inferring the schema.