Cerin Amroth — ML, JWST & Security Research

Independent research in machine learning, astronomy and software security: run large models on accessible hardware, search public telescope images, and test what software promises against what it does.

ML systems — quantized Mixture-of-Experts training and inference across GPU memory, host memory, and NVMe

experts4bit-qlora Train and serve Mixture-of-Experts models that do not fit in VRAM: fused 4-bit experts, QLoRA, CPU/NVMe offload, fast inference on consumer NVIDIA GPUs. v0.48.0 · Install, supported paths and evidence → grouped-nf4-gemm Triton kernels for 4-bit MoE inference: grouped NF4/MXFP4 GEMM, INT4 GEMV, FP8 paged attention, CPU/NVMe expert streaming. v0.41.0 · Install, supported paths and evidence → OpenVINO Upstream fixes keeping Intel-GPU LLM inference off the CPU-fallback path — three merged.

GitHub: experts4bit-qlora · PyPI · problems it solves · GitHub: grouped-nf4-gemm · PyPI · problems it solves

Start with the problem

Problem	Route
My MoE still OOMs after `load_in_4bit=True`	experts4bit-qlora
I need to QLoRA-train fused MoE experts	experts4bit-qlora
My model is larger than GPU or host memory	e4b residency paths
Dequantizing every expert before GEMM is too slow	grouped-nf4-gemm
I need native MXFP4 or INT4 expert compute	grouped-nf4-gemm
I want to inspect the evidence methodology	Receipts-driven engineering

Observational — JWST imaging, measured rather than benchmarked

Little red dots — an ongoing search Map published little red dots, explore an ongoing search of public JWST images, and follow individual sources through real telescope observations.

Security research — good-faith vulnerability research and coordinated disclosure

Vulnerability research Good-faith research & coordinated disclosure. Published disclosures ZisK zkVM: a soundness bug in signed remainder, fixed upstream. Vendor advisory GHSA-qjgg-fcj3-p9x3, rated Critical.

Contact: jordan@cerinamroth.com · security reports: security@cerinamroth.com · security.txt · consulting: jordananderson.work