Koorous Vargha developed the Venice RocksDB integration for the facebook/rocksdb repository, enabling Venice to use RocksDB as the storage backend for LinkedIn’s machine learning feature store. This work established a scalable, low-latency storage path for ML feature retrieval, supporting production recommendation pipelines. Koorous focused on data storage and database management, implementing the integration pathway and documenting the process in Markdown. The project addressed the need for efficient, production-grade feature serving by connecting Venice with RocksDB, allowing seamless retrieval of ML features. The depth of the work lies in bridging two complex systems to streamline LinkedIn’s recommendation infrastructure for machine learning.