To build this, you need to first check out the latest spark source https://github.com/apache/spark and run sbt/sbt publish-local. This will put spark in maven's local repository.
You will also want to build this version of spark with ./make-distribution.sh --tgz
Then from this repository, run mvn package to create avro-0.1.jar.
To start up pyspark with with this library, run SPARK_CLASSPATH=/tmp/avro-0.1.jar bin/pyspark from the distribution you created above.
To load in an Avro file in pyspark use:
avroRdd = sc.newAPIHadoopFile("/tmp/data.avro",
"org.apache.avro.mapreduce.AvroKeyInputFormat",
"org.apache.avro.mapred.AvroKey",
"org.apache.hadoop.io.NullWritable",
keyConverter="example.AvroGenericConverter")