I study how to make large-model inference smaller, faster, and still trustworthy: train the compact head, prune the experts, assign the precision—then measure the claim in a runtime built to expose it. Everything here is a working record; the results are the signal.
Systems software engineer, AWS ElastiCache · Maintainer, Valkey GLIDE · Founder, agent-shEach page separates the question, protocol, evidence, negative results, and what remains unresolved. PDFs arrive when the experiment is complete—not before.
Can a compact, separately trained MTP head replace the inherited full-width block and 248k-row output projection without surrendering the acceptance that makes speculation useful?
A study of compression order: remove experts with router-weighted activation evidence, heal the surviving function, then assign low precision—rather than asking quantization or pruning to carry the full loss alone.
I keep the failed arms because they define the research more honestly than a clean success narrative does.
The current StudentSV arm reaches 0.527 code slot-0 acceptance against a 0.675 teacher at roughly one eighth of the estimated draft FLOPs. It is not yet at the success threshold.
It did not. Across forced replay, generation-distribution replay, and seeded sampled runs, two clean quantization arms showed no acceptance loss versus BF16. The original healing premise was closed.
The next MoE study asks whether routing frequency, activation impact, and spill cost can predict which surviving experts deserve higher precision—and whether the policy survives a generative evaluation.
Research claims about inference depend on loaders, kernels, caches, evaluation code, and hardware behavior. I build enough of that stack to know what the number includes.
A from-scratch Rust + CUDA inference engine used as an experimental instrument for MTP, sampled speculative decoding, NVFP4, MoE spill, and exactness gates on Blackwell.
public · active ↗ 02 / study hqmtpThe research harness for quantized-trunk MTP agreement and compact StudentSV heads: extraction, training, replay evaluation, corpus controls, and negative-result ledgers.
working paper 03 / compress REAP × quantExpert saliency, pruning, full-function healing, low-bit simulation, and the loader work needed to carry REAP50 artifacts into a constrained inference runtime.
working paper 04 / migrate CRIU live migrationCopy-on-write live migration of a 200 GB loaded in-memory datastore with under 50 ms of freeze: userfaultfd page prefetch, priority-ordered VMA transfer, and lazy-page lifecycle fixes.
research fork ↗ 05 / systems Valkey GLIDEProduction systems work at a different scale: an official multi-language Valkey client. 131 authored pull requests merged and 410 reviewed—reliability, releases, and the support that starts after an implementation works once.
maintainer ↗ 06 / desktop ChatGPT desktop for LinuxBringing OpenAI's desktop app to Linux: Wayland input, accessibility-tree automation, packaging across five formats, and the updater plumbing on a 2k-star community project.
collaborator ↗ 07 / tools agent-shOpen-source agent tooling organization: configuration linting, code review, repository intelligence, and isolated computer-use environments.
founder ↗The paper pages are formal and evidence-heavy. The notes are where systems lessons, design arguments, and open-source experience can breathe.
What streaming, spending, waiting, and long-lived work revealed about the queue's model of execution.
Why silent failures in skills, hooks, memory, MCP, and instruction files needed source-backed validation.
I am an independent ML researcher and a systems software engineer at AWS ElastiCache. My route into research runs through implementation: reproduce the baseline, expose the hidden assumption, train the missing comparison, and let the result change the plan.
I study computer science at The Open University of Israel while maintaining open-source systems and running research outside a traditional lab. That path has made method unusually important to me. Claims need matched baselines, saved artifacts, explicit noise floors, and a visible record of negative results.
The research is trained, not only measured: compact MTP draft heads with their own vocabularies, healed pruned MoEs, a fine-tuned ColBERTv2 retriever with a LoRA relevance judge to score it. Before focusing on efficient inference I worked deeply in datastores, language clients, queues, retrieval, and developer tools; I still maintain Valkey GLIDE, stay active across the valkey-io ecosystem, and help build desktop developer tools, and that record—review, releases, support—runs alongside the studies. Model research becomes real software surprisingly quickly.
I also volunteer as a mentor with Yotzim LaShinui, helping a junior developer from a formerly ultra-Orthodox background find their way into software.