Linked Presentation: DINT: Fast In-Kernel Distributed Transactions with eBPFTHC: Accelerating Distributed Deep Learning Using Tensor Homomorphic Compression