Currently, he is working on an efficient distributed NoSQL data store optimized for NVMe storage (Flash- and 3DXP-based SSDs), which has been released as an experimental service in the IBM Cloud. Recently, he's interested in accelerating large scale machine learning systems, and is involved in the Snap ML project.

Nikolas has also worked on distributed flash caching, flash controller optimizations, data deduplication, and storage class memories. Projects that he have been involved with include the IBM FlashSystem, the IBM Easy Tier Server for DS8000 storage, the SALSA unified software storage stack for SSDs and SMR drives, and the iostash flash caching open-source project.