Filip Makraduli
Applied AI engineer and researcher at Superlinked, where he builds infrastructure for self-hosted small-model inference, retrieval, and agent workflows. He co-authored the FlashNorm research on faster RMSNorm execution and speaks publicly on GPU inference and CUDA optimization.
FlashNorm Folds RMSNorm Into Projection Weights to Cut Decode Time
Filip Makraduli argues that RMSNorm’s inference cost comes less from arithmetic than from repeated GPU launches, memory movement and dependencies that hold up projection matmuls. His FlashNorm approach exactly rewrites the norm–projection sequence by folding learned gains into weights, deferring the RMS scalar division until after the matmul, and in some architectures removing a redundant normalization. Weight folding can speed existing checkpoints now, he says, while the larger benefit from deferred normalization depends on fused kernels and correct CUDA-stream synchronization.
Small-Model Inference Needs Infrastructure Beyond Model Servers
Filip Makraduli of Superlinked argues that the hard part of small-model inference is no longer simply serving a model, but operating many embeddings, rerankers, extractors and multimodal models efficiently in production. In his account, conventional one-model-per-container deployments waste GPU capacity and leave teams to rebuild routing, autoscaling, monitoring, hot-swapping and eviction themselves. Superlinked’s SIE is presented as an open-source attempt to provide that missing infrastructure layer for AI search and document-processing workloads.