The talk is about open CSD, a computational storage emulation platform. The speaker explains what computational storage is and why it is needed in today's data-driven society. They discuss the challenges posed by high data and throughput requirements on storage interfaces and technologies. Computational storage devices, which have their own CPU and memory, can be used to offload computations from the host processor, reducing data movement and improving energy efficiency. The speaker presents OpenCSD, an open-source solution that aims to standardize the API and file system for computational storage. They explain the key components of OpenCSD, including a log-structured file system, module interface, and the use of zone namespaces in flash SSDs. They also discuss the use of EBPF for defining system calls and making the solution architecture-independent. The speaker presents a demo of the system using Shannon entropy as an example use case. They discuss the limitations of the prototype and suggest possible solutions for future development. The talk concludes with an invitation to try OpenCSD and references to related work and resources for further reading.