39-day current streak·88-day longest streak
Hey, I'm Avi 👋 I like machines close to the metal and results you can measure. By day I build in-memory data systems at AWS ElastiCache; the rest of the…
Hey, I'm Avi 👋
I like machines close to the metal and results you can measure. By day I build in-memory data systems at AWS ElastiCache; the rest of the time I train and compress small fast models, write CUDA, and maintain open-source tools people actually run.
Research notes and working papers live at avifenesh.ai. Issues, questions, and counterexamples are always welcome.
ML research
- bw24 — from-scratch Rust + CUDA inference engine for one Blackwell GPU. Bit-exact by construction; runs four model families above llama.cpp on the same rig, plain and speculative decode both.
- hqmtp — MTP draft-head lab. A 212M drafter distilled to 85% of its teacher's chain acceptance at ~13% of the draft compute. The negative results stay in the ledger.
- Working papers: small-vocabulary MTP heads · prune, heal, quantize — methods, failed arms, and evidence in the open.
- In review upstream: NVFP4 quantization at llama.cpp, hybrid-KV loads at vLLM and LMCache.
Maintaining
- Valkey GLIDE — the official multi-language Valkey/Redis client. Rust core, Java/JNI, Node/N-API; 131 PRs authored, 410 reviewed, plus releases and the support that starts after the code works.
- valkey-skills — official AI skills for the Valkey ecosystem. Started it, maintain it.
- agent-sh — my org: an ecosystem of tools for agent-assisted development, working across Claude Code, Codex, OpenCode, Cursor, and Kiro.
- glide-mq — Node.js queue on Valkey Streams with a Rust N-API core, plus adapters for Hono, Fastify, Hapi, NestJS and a dashboard.
- Also around: RustOwl (runtime, memory, and CI work) and ChatGPT Desktop for Linux (Wayland input, accessibility, packaging).
Systems
- Valkey — core contributor; sync-from-replica replication in review. On the side: CRIU copy-on-write live-migration research — under 50 ms of freeze while migrating a 200 GB loaded instance.
- FlowFabric — durable-execution engine in Rust for Valkey, Postgres, and SQLite: lease-safe workers, waitpoints, budgets.
- layout-audit — DWARF memory-layout analysis: padding, layout diffs, size budgets for C/C++/Rust/Go.
- scrump — format-aware secret scrubber for binary capture artifacts: perf.data, core dumps, nsys traces, JFR.
- ocaml-valkey — OCaml 5 + Eio Valkey client, published on opam.
Agents
- agnix — linter and language server for AI agent configs: 432 rules with autofixes, a GitHub Action, an MCP server, and editor plugins.
- computer-use-linux / agent-workspace-linux — Linux desktop control over MCP, and isolated agent-owned desktops so an agent never has to touch your real machine.
- eigen — local agent daemon in Go: resumable sessions, multi-provider routing, memory, observability.
Elsewhere
- Talks: Inside Valkey GLIDE on the AWS Developers Podcast · Glide into resiliency on Let's Talk About Data
- Writing: avifenesh.ai/writing · answering on Stack Overflow
- 📫 [email protected] · LinkedIn · X
-
bw24 ★ PINNED
From-scratch Rust+CUDA inference engine, bit-exact by construction — NVFP4, MoE, MTP speculative decoding, tuned against measured limits of one RTX 5090 Laptop (sm_120a).
Rust ★ 266 18h agoExplain → -
hqmtp ★ PINNED
MTP draft-head research harness for Qwen3.5-9B: StudentSV compact drafter, chain distillation to 85% of teacher acceptance at 13% of draft compute, and the negative-result ledgers behind it. Measured with bw24.
Python ★ 1 3d agoExplain → -
glide-mq ★ PINNED
High-performance message queue for Node.js — Valkey/Redis Streams with Rust-native NAPI bindings
TypeScript ★ 92 12d agoExplain → -
claucode.nvim
A Neovim bridge plugin for Claude Code CLI, providing seamless integration between your editor and Claude's AI capabilities.
Lua ★ 23 2mo agoExplain → -
glidemq-hono
Hono middleware for glide-mq — REST API + SSE events for queue management
TypeScript ★ 13 11d agoExplain → -
ocaml-valkey
Modern Valkey client for OCaml 5 + Eio (RESP3-only).
OCaml ★ 10 1mo agoExplain → -
FlowFabric
Valkey-native execution engine for AI and workflow systems
Rust ★ 5 12d agoExplain → -
glide-distributed-lock
A robust distributed lock implementation for Valkey cluster
TypeScript ★ 5 12d agoExplain → -
layout-audit
GitHub Action & CLI to analyze binary memory layouts: detect padding, compare diffs, enforce budgets. Parses DWARF debug info for C/C++/Rust/Go.
Rust ★ 5 6mo agoExplain → -
glidemq-nestjs
NestJS module for glide-mq - decorators, dependency injection, and lifecycle management for queues and workers
TypeScript ★ 5 1mo agoExplain → -
codex-desktop-linux ⑂
Run OpenAI Codex Desktop on Linux - automated installer
JavaScript ★ 4 1h agoExplain → -
ferrings
Node.js library for faster, lower-latency TCP — powered by Linux io_uring. Up to 2.5x throughput and fewer syscalls than native Node networking.
JavaScript ★ 4 6h agoExplain → -
opam-repository ⑂
Main public package repository for opam, the source package manager of OCaml.
★ 3 3d agoExplain → -
revuto
Local, supplier-agnostic, repo-agnostic autonomous PR reviewer that learns from maintainer feedback into graduated skills (any OpenAI-compatible model; Obsidian vault; SurrealDB/SQLite).
TypeScript ★ 3 3d agoExplain → -
sxc
Local corpus retrieval over your own text: BM25 + SPLADE v3 + ColBERTv2, with synthetic eval and an optional relevance judge
Python ★ 3 16d agoExplain → -
glidemq-hapi
Hapi plugin for glide-mq - queue management REST API and SSE events
TypeScript ★ 3 12d agoExplain → -
tools
harness - Agent tools (read, write, grep, glob, bash, webfetch, lsp, skill) with Rust and TS parity. Designed for real LLMs usage.
TypeScript ★ 3 13d agoExplain → -
glidemq-dashboard
Web dashboard for glide-mq — real-time queue monitoring and management
HTML ★ 3 13d agoExplain → -
glidemq-fastify
Fastify plugin for glide-mq queue management
TypeScript ★ 3 1mo agoExplain → -
glide-mq.dev
Documentation site for glide-mq - High-performance message queue for Node.js
Python ★ 3 2mo agoExplain → -
eigen
Opinionated, terminal-first local coding agent for Go/Linux: one daemon, many resumable sessions, multi-provider routing (Bedrock/Claude/OpenAI-compatible/local), memory, dreaming, a proactive feed, and observability. Built for myself.
Go ★ 2 5d agoExplain → -
algo-zoo
No description.
Rust ★ 2 11d agoExplain → -
glidemq-examples
Examples and integrations for glide-mq - message queue for Node.js on Valkey/Redis
★ 2 1mo agoExplain → -
leetcode-exporter
No description.
JavaScript ★ 2 7mo agoExplain → -
30-Colorful-Days-of-Vim
A month of one challenge a day for vim learning - while the chllenges written in a colorful set of languages, all you need is to follow the vim challenge of the day.
Shell ★ 2 4mo agoExplain → -
extensions ⑂
Extensions for the Zed editor
JavaScript ★ 1 11h agoExplain → -
hermes-agent ⑂
The agent that grows with you
Python ★ 1 3d agoExplain → -
fastify ⑂
Fast and low overhead web framework, for Node.js
JavaScript ★ 1 3d agoExplain → -
rustowl ⑂
Visualize Ownership and Lifetimes in Rust
Rust ★ 1 3d agoExplain → -
valkey ⑂
A flexible distributed key-value database that is optimized for caching and other realtime workloads.
C ★ 1 3d agoExplain → -
valkey-glide ⑂
No description.
Java ★ 1 3d agoExplain → -
avifenesh
No description.
★ 1 3d agoExplain → -
speedkey
No description.
Rust ★ 1 22h agoExplain → -
scrump
Fast, format-aware secret scrubber for binary capture artifacts (perf.data, nsys-rep, ELF core, hprof, JFR, pcap, SQLite, tar/zip).
Rust ★ 1 11d agoExplain → -
valkey-skills ⑂
Domain-specific AI skills for working with Valkey - reusable knowledge packs that help agents and developers write correct, idiomatic, high-quality code against the Valkey ecosystem.
★ 1 11d agoExplain → -
glide-consumer-supplier-queue
A simple yet powerful task queue implementation using Valkey streams and the GLIDE client library. Available in both Python and Node.js.
TypeScript ★ 1 13d agoExplain → -
llama.cpp ⑂
LLM inference in C/C++
C++ ★ 1 15d agoExplain → -
tuicr ⑂
A human-in-the-loop code review TUI for AI-generated changes.
Rust ★ 1 1mo agoExplain → -
GFit
GFit - Personal Trainer Website for Gad Panesh
HTML ★ 1 2mo agoExplain → -
balance-beacon ▣
No description.
TypeScript ★ 1 3mo agoExplain → -
A-bit-racy ▣
a cute car game created by python code (pygame)
Python ★ 1 6y agoExplain → -
preprocessor
Semi preprocessor that cleans the file from comments and includes the local "include" files
C ★ 1 5y agoExplain → -
assembler_openU
Final project of Systems programming lab course
C ★ 1 5y agoExplain → -
efficiencyOfCountDiffVall
Count the number of oppratations, and check the efficiency of 7 different algorithms, counting the amount of unique values in a given array
Java ★ 1 5y agoExplain → -
assembly-ex
assembly exercises
Assembly ★ 1 3y agoExplain → -
mouse-mover
A simple script to mimic human mouse movement so the manager of my girlfriend wont be mad if she takes more than 30 min launch break
Python ★ 1 1y agoExplain → -
my-solution-to-code-problem
Sharping my solving problem abilities, write solution on a white-board and then code it to valid my answers
JavaScript ★ 1 1y agoExplain → -
sortLogFiles
Algorithm for sorting multiple sorted log-files into one sorted log-file base on time-stamp
Python ★ 1 1y agoExplain → -
node-rate-limiter-flexible ⑂
Atomic counters and rate limiting tools. Limit resource access at any scale.
JavaScript ★ 1 10mo agoExplain → -
into-engineering
No description.
C ★ 1 9mo agoExplain → -
df-regexp-extract
No description.
Rust ★ 1 9mo agoExplain → -
efficiency-game
No description.
HTML ★ 1 8mo agoExplain → -
multithreading-workshop
No description.
C ★ 1 8mo agoExplain → -
valkey-admin ⑂
No description.
★ 1 5mo agoExplain → -
zed ⑂
Code at the speed of thought – Zed is a high-performance, multiplayer code editor from the creators of Atom and Tree-sitter.
★ 1 4mo agoExplain → -
hapi.dev ⑂
The hapi.dev developer portal
★ 1 1mo agoExplain → -
signbridge
Real-time ASL signing overlay for Telegram calls — open source Android app for deaf users
Python ★ 0 22h agoExplain → -
vllm ⑂
A high-throughput and memory-efficient inference and serving engine for LLMs
Python ★ 0 3d agoExplain → -
LMCache ⑂
Supercharge Your LLM with the Fastest KV Cache Layer
Python ★ 0 18h agoExplain → -
pearl ⑂
Monorepo for the Pearl network 🐚
Go ★ 0 3d agoExplain → -
atrium
Zero-dependency, single-machine life dashboard — GitHub tasks, local AI agents, system health, email/calendar, subscriptions — with a global quiet switch, a pluggable collector model, and an MCP server. Loopback-only by design.
TypeScript ★ 0 5d agoExplain → -
collide
No description.
TypeScript ★ 0 12d agoExplain → -
floe
Architectural PR review for TypeScript — turns a PR into flows and asks three questions per flow: does it deliver what's claimed, is there proof, how much harder did it make the code to navigate.
Rust ★ 0 13d agoExplain → -
gemma-expert-atlas
No description.
Python ★ 0 25d agoExplain → -
tfqol
Tools for quality of life — small, focused Linux utilities
Python ★ 0 1mo agoExplain → -
napi-rs ⑂
A framework for building compiled Node.js add-ons in Rust via Node-API
★ 0 1mo agoExplain → -
ort ⑂
Fast ML inference & training for ONNX models in Rust
★ 0 1mo agoExplain → -
team-tmux
Run a 1-manager + 3-worker AI agent team in tmux. File-based message queue with atomic delivery.
Shell ★ 0 3mo agoExplain → -
awesome-hapi ⑂
A curated list of delightful HapiJS packages & resources
★ 0 4mo agoExplain →
No repos match these filters.