Skip to content
View hizrianraz's full-sized avatar
🚀
To Infinity and Beyond!
🚀
To Infinity and Beyond!

Highlights

  • Pro

Block or report hizrianraz

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
.github/profile/README.md
    __  ______
   / / / / __ \
  / /_/ / /_/ /
 / __  / _, _/
/_/ /_/_/ |_|

Hizrian Raz

Philosopher · Wanderer · Neurodivergent · Founder

Building Ainfera — an AI-Native model factory.
On the side: personal measured packs for one box. Receipts over hype.


Ainfera LinkedIn X Hugging Face GitHub Website


Typing SVG


About

I'm a founder who cares more about what actually runs than what trends.

  • I build Ainfera — an AI-Native model factory (company work lives under @ainfera-ai).
  • I publish personal, unaffiliated measurement packs so one DGX Spark stays reproducible.
  • I optimize for evidence, clarity, and long-horizon craft — not launch theater.

Packs here are not company IP, not finetunes, and not invented numbers.


Skills

Domain What I actually do
Model systems Serve paths, quant honesty, agentic runbooks, eval receipts
Inference craft llama.cpp / GGUF pins, FP8 quality tracks, single-node MoE fit
Agent loops Tool-use harnesses, format/routing smoke, hermes-style checks
Evidence discipline Digests, freeze clocks, SAQS claim binder, smoke ≠ headline
Product eng TypeScript surfaces, Python toolchains, shell pack automation
Founder ops Spec → measure → ship, ADHD-aware execution, conviction over consensus

Python TypeScript Shell PyTorch Hugging Face llama.cpp GGUF CUDA vLLM Docker Linux macOS Git GitHub VS Code


Focus right now

Personal measured stack on one DGX Spark (GB10 · 128 GB).

Scripts · pins · cards · eval receipts.
Not finetunes. Not company IP. Not stretch claims.

# Pack Role (honest) Status
01 Laguna-S-2.1 · HF Long-horizon repo maintenance · official Q4_K_M Measured day-0
02 Qwen3-Coder-Next · HF Interactive throughput · quality FP8 Day-0 path
03 DeepSeek-V4-Flash REAP25 · HF Long-context investigation · REAP25/Pulsar Experimental · no hero

Laguna-S Spark Qwen3-Coder-Next DeepSeek REAP25

Laguna-XS Mac Qwen 30B Mac HF models

Also on Mac (not Spark heroes):
Laguna-XS-2.1 · Qwen3-Coder-30B-A3B

Size · pins · windows

Window (WIB)
List target 2026-08-03 12:00 · freeze target 2026-08-02 18:00
Both clocks stay blocked until local HOLD lifts.

Hero CTAs (sequential)
Laguna Aug 3 20:00 → Qwen Aug 4 20:00 → DeepSeek null (NO_HERO default)

Laguna day-0 measured artifact
hizrianraz/Laguna-S-2.1-GGUF · laguna-s-2.1-Q4_K_M.gguf
sha256 a8b55c75714ea73fd90ec85de5defdc0b8d88ca0ad2108343cdd8fc22f7583e4
engine pin 04b2b72 (poolside/llama.cpp-laguna) · measure tip bf82eab
format/routing smoke 40/40 · hermes 27/27 · ~21.47 t/s gen128
not long-horizon agent reliability proof · DFlash/NVFP4 not day-0 flagship

Upstream / quality refs (not day-0 serve authority)
poolside/Laguna-S-2.1-NVFP4 · Qwen/Qwen3-Coder-Next-FP8 · twaggs88 REAP25 DSpark GGUF · bases Laguna-S-2.1 · Qwen3-Coder-Next · DeepSeek-V4-Flash


Operating system

measure  →  pin  →  publish receipts  →  refuse the stretch claim
Principle In practice
Conviction over consensus One box. Real digests. Reproducible scripts.
Verify the premise Smoke ≠ headline. Verifier ≠ gate clearance.
Honest labels Experimental stays experimental until dated measure lands.
Integrity is structural Claim binder ships with the pack — not a blog afterthought.

Public claim binder for every Spark pack:
SPARK_AGENTIC_QUANT_STANDARD.md (SAQS)

Honesty locks — always on
  • diy_gguf = false — Laguna GGUF mirror hosts official Poolside bytes only
  • public_promo_before_launch = false
  • smoke ≠ agent headline · evidence-bind ≠ gate
  • Qwen day-0 forte = FP8 quality (not NVFP4 speed)
  • DeepSeek day-0 = scaffold / experimental · NO_HERO default
  • DSpark ≠ DGX Spark · peer GGUF ≠ full official DeepSeek V4 Flash (~155 GiB no-fit on one Spark)
  • Packs are not Ainfera product surfaces, company eval, sales SKUs, or open finetunes

Reproduce a pack

1  read README + INSTALL.yaml
2  pull the named upstream  (or Laguna official GGUF + SHA256SUMS)
3  run scripts/serve_*.sh   on one DGX Spark
4  diff results/            — smoke ≠ headline

Surface

Static links only — no third-party stats widgets (those CDN pins break and show empty images).

Public repos Ainfera org LinkedIn HF X

Track Home
Personal packs github.com/hizrianraz
Company ainfera.ai · @ainfera-ai
LinkedIn linkedin.com/in/hizrian-raz
Models huggingface.co/hizrianraz
Profile binder SPARK_AGENTIC_QUANT_STANDARD.md

Elsewhere

Company ainfera.ai · @ainfera-ai
LinkedIn linkedin.com/in/hizrian-raz
Models huggingface.co/hizrianraz
HF profile space huggingface.co/hizrianraz/hizrianraz
Pack questions open a discussion on the relevant HF repo

Thanks for reading this far.

Personal · unaffiliated measurement surface · not Ainfera product IP Polished 2026-07-30 · Laguna measure authority bf82eab

Pinned Loading

  1. ainfera-ai/ainfera-evals ainfera-ai/ainfera-evals Public

    VAC/$ + certificates (hero public product — private until launch)

    Python

  2. Laguna-S-2.1-Spark-Agentic Laguna-S-2.1-Spark-Agentic Public

    Personal measured Laguna-S-2.1 pack · Spark-only full MoE · Q4 headline · promo 2026-08-03 12:00 WIB

    Python

  3. Qwen3-Coder-Next-Spark-Agentic Qwen3-Coder-Next-Spark-Agentic Public

    Spark stand-behind agentic pack — Qwen3-Coder-Next official Q4_K_M (diy_gguf=false). Pull/serve scripts + smoke harness. Companion to Laguna S+XS Aug 3 set.

    Shell

  4. DeepSeek-V4-Flash-Spark-Agentic DeepSeek-V4-Flash-Spark-Agentic Public

    Card-only hold scaffold for DeepSeek-V4-Flash on DGX Spark (personal, unmeasured)

    Shell