Skip to content
View paperniuk's full-sized avatar
30 followers · 29 following

Achievements

Achievement: Pull SharkAchievement: Starstruckx2Achievement: Quickdraw

Achievements

Achievement: Pull SharkAchievement: Starstruckx2Achievement: Quickdraw

Highlights

  • Pro

Block or report paperniuk

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. 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
paperniuk/README.md

Erik Paperniuk (韓勝利)

System integration and automation engineer based in Taipei, Taiwan.

I build backend services and turn fragmented manual processes into software: API integrations, ERP / CRM automation, data pipelines, and applied ML systems.

Website · LinkedIn · Repositories

What I work on

  • Backend services and API integration
  • ERP / CRM automation — Frappe, ERPNext, Odoo
  • Private LLM deployments and applied AI
  • GPU kernels and local LLM inference on Apple Silicon (Metal)
  • Computer vision and applied machine learning

Stack

  • Languages — Python, C++, TypeScript, JavaScript, Metal Shading Language
  • Backend — FastAPI, Django, Flask
  • Data — PostgreSQL, MySQL
  • ML — PyTorch
  • Infrastructure — Docker, Linux, Git, Cloudflare
  • Business systems — Frappe, ERPNext, Odoo

Selected projects

Splash for M1/M2 Macs Port of the Splash LLM inference engine to M1/M2 with custom Metal kernels: Qwen3.8-27B decode about 2x and prefill 2.3-2.7x faster than the stock kernels on an M1 Max, prebuilt one-line install.

LM Studio Prism Bonsai Makes LM Studio load Ternary Bonsai 2 GGUF models (PQ2_0 / PTQ1_0) through Prism's llama.cpp fork, with a reasoning-effort selector. Linux and Windows, double-click install.

oMLX Prism Hadamard Makes oMLX run Prism's Hadamard Qwen3.5 packs (Ternary Bonsai 2, a 27B model in 8.4 GB), plus a guide to adding unrecognised architectures to mlx-vlm.

Chat Data Exporter Desktop app for extracting and structuring public contact information from large website collections.

Moodle PWA Plugin Enables PWA installation for Moodle while preserving site branding.

PrepPilot AI Mobile app for AI-assisted exam-prep workflows.

Research

Deepfake and image-authenticity detection · physics-informed neural networks · robust model deployment.

Open to collaboration on backend, automation, and applied AI work.

Pinned Loading

  1. splash splash Public

    Forked from incoai/splash

    Unofficial M1/M2 port of Inco's Splash (upstream is M3+ only): Metal kernels written for Apple7/8 GPUs, prebuilt releases.

    C++ 219 18

  2. ds4 ds4 Public

    Forked from antirez/ds4

    Fork of antirez/ds4 for Apple Silicon, tuned mostly for M1/M2: Qwen3.8-Flash-Next with 262K context on 64 GB and up, 35 tok/s on an M1 Max, MTP and vision, one-line install.

    C 38 3

  3. lmstudio-prism-bonsai lmstudio-prism-bonsai Public

    Make LM Studio load Ternary Bonsai 2 GGUF (PQ2_0 / PTQ1_0) via Prism's llama.cpp fork, with reasoning-effort selector. Linux + Windows, double-click install.

    PowerShell 32

  4. chat-data-exporter chat-data-exporter Public

    A fast, accurate PyQt6 desktop application for extracting WhatsApp and Telegram contact information from thousands of websites.

    Python

  5. moodle-pwa-plugin moodle-pwa-plugin Public

    Moodle local plugin that enables PWA install via Chrome - uses your site's logo automatically.

    PHP 1

  6. AkmalAgzamov1/YZU-judge AkmalAgzamov1/YZU-judge Public

    TypeScript 1 1