2026-06-16
Building a from-scratch differentiable spectral-segmentation library in PyTorch — hand-coding learned-metric affinity kernels, a cyclic Jacobi + Lanczos eigensolver, and a degeneracy-robust differentiable backward with Lorentzian-broadened eigenvalue-gap gradients stable where eigenvalues collapse at segment boundaries.
spectral-graph-theorypytorchcomputer-visionsegmentationeigensolvermathematics
2026-06-16
How I built a Redis-wire-compatible (RESP) vector memory engine in Rust with zero dependencies beyond libc — hand-rolling an HNSW index, SIMD distance kernels in NEON and AVX2+FMA with runtime CPU dispatch, and a single-threaded event-driven I/O core over raw kqueue/epoll.
rustsystemsvector-searchHNSWSIMDredis
2026-06-03
Why norepinephrine does not reconfigure brain circuits through neurons but through astrocytes, how the tripartite synapse implements a biological diffusion noise schedule, and what every AI architecture is missing.
astrocytesneurosciencetripartite-synapsenorepinephrineDAHNdiffusion-memory
2026-02-19
Inheritance, polymorphism, encapsulation, abstraction, composition, and the diamond problem , explained through one of my favs Clash Royale.
OOPjavagame-designprogrammingclash-royale
2026-02-18
Why intelligence is not continuation but convergence & exploring energy based models, dynamical memory fields, differential decay, and the geometry of reasoning.
energy-modelsfree-energydynamical-systemsAGImemory
2026-02-15
Exploring the deep connection between diffusion on graphs, the graph Laplacian, and how the brain consolidates memory through relaxation, inhibition, and cortical smoothing.
neurosciencespectral-graph-theorymemorymathematics