NG

Xuan-Son Nguyen

@ngxson

The sky is blue. The sun is yellow. Here we go. There and back again.

France

Website

Joined May 2014

1039 Followers

39 Following

AI Insights & Achievements

70/ 100

Quality

Avg 74.0 stars per repo.

Growth

16 new projects with 5823 stars.

Consistency

72 day streak.

Profile README

Hacking things on llama.cpp [ Interested in my works? Buy me a ☕ or Sponsor me 🩷 ]

New jinja template engine: https://github.com/ggml-org/llama.cpp/pull/18462
Adding MiMo-V2-Flash support: https://github.com/ggml-org/llama.cpp/pull/18328
Adding GLM-4V support: https://github.com/ggml-org/llama.cpp/pull/18042
The brand new llama-cli: https://github.com/ggml-org/llama.cpp/pull/17824
Serving multiple models in parallel via llama-server: https://github.com/ggml-org/llama.cpp/pull/17470
Shipping Ministral 3, Devstral 2 (partnership with Mistral): https://github.com/ggml-org/llama.cpp/pull/17644
Shipping GPT-OSS (partnership with OpenAI and GGML team): https://github.com/ggml-org/llama.cpp/pull/15091
Bringing vision support to llama-server: https://github.com/ggml-org/llama.cpp/pull/12898 (on hackernews)

and more...

(Big) refactoring vision support in llama.cpp, introducing libmtmd: https://github.com/ggml-org/llama.cpp/pull/12849
Support various vision models: Pixtral, SmolVLM, etc (check out this viral demo)
Gemma 3 Vision support (partnership with Google): https://github.com/ggml-org/llama.cpp/pull/12343
WASM speed improvement: https://github.com/ggml-org/llama.cpp/pull/11453 (also checkout wllama and hackernews post)
Refactor argument parser, for a better CLI UX: https://github.com/ggml-org/llama.cpp/pull/9308
Revamp llama.cpp Web UI (currently deprecated, but was the precursor for the modern version): https://github.com/ggml-org/llama.cpp/pull/10175
llama.cpp <> Hugging Face inference endpoint integration: read docs
Hot-swapping LoRA adapters: https://github.com/ggml-org/llama.cpp/pull/8332
Control vector generator: https://github.com/ggml-org/llama.cpp/pull/7514
Initial chat template support: https://github.com/ggml-org/llama.cpp/pull/5538

Organizations

HU

Public Repos

159

Total Stars

7395

Total Forks

1145

Public Gists

26

Top Languages

Based on primary language of repositories

Recent Activity

Contribution Graph

Activity Timeline

Commits and contributions grouped by day, week, or month.

High productivity periods

Most active day: January 27th, 2026 • 16 contributions

Most active week: Dec 1–Dec 7 • 75 contributions

Most active month: May 2025 • 558 contributions

Pushed to master at ngxson/llama.cpp

February 1st, 2026 10:26 AM

opened pull request in crosspoint-reader/crosspoint-reader

January 31st, 2026 10:02 PM

Created branch xsn/apps in ngxson/crosspoint-reader

January 31st, 2026 9:01 PM

Pushed to xsn/mqjs at ngxson/pluspoint-reader

January 31st, 2026 5:18 PM

Pushed to master at ngxson/llama.cpp

January 31st, 2026 4:02 PM

Pushed to xsn/mqjs at ngxson/pluspoint-reader

January 31st, 2026 2:27 PM

Pushed to master at ngxson/llama.cpp

January 31st, 2026 5:17 AM

Pushed to master at ngxson/llama.cpp

January 31st, 2026 3:41 AM

IssueCommentEvent on ggml-org/llama.cpp

January 31st, 2026 12:23 AM

IssueCommentEvent on ggml-org/llama.cpp

January 30th, 2026 11:23 PM

GitHub Analytics

Xuan-Son Nguyen

Quality

Growth

Consistency