2016-2018

• Built and rearchitected Moat’s realtime data pipeline, which handles 80 billion messages a day using AWS Kinesis, VoltDB, and stream processing applications based on Apache Flink and Spark.
• Built a distributed datastore and query service that uses AWS Lambda to perform parallel queries of flat files stored in S3 buckets. It’s a snappy, cheap way to store and query archival stats data.
• Migrated stats databases to use more effective schema, decreasing complexity and latency.