amazon web services – Kafka process batch data per partition

I have a Kafka topic with thousands of partitions (each partition containing data from a particular user) and need to process the data efficiently. My goal is to trigger an AWS Lambda function when 20 messages accumulate in a partition (essentially 20 messages for a user). Is it possible to configure Lambda to process messages in batches per partition, rather than usual batching at the topic level? If not what should be that possible approach to process the data in batches for each user?