VJML / LAB

Lab

Experiments, notes and systems under investigation.

Things here may work, fail, change or eventually become projects.

01

Local AI

Local inference, retrieval, embeddings and generative systems.

2026.09 ACTIVE

LLM inference on a legacy GPU

llama.cpp / CUDA / Pascal / LLM
2026.10 ACTIVE

Building a local RAG

RAG / Embeddings / Vector Search / LLM
2026.10 ACTIVE

Image generation on Apple Silicon

Image Generation / Apple Silicon / Local Inference / Benchmarking
2026.10 ACTIVE

Local image generation pipeline

Image Generation / Prompt Engineering / Automation / Local Inference
02

Architectures

ARM, RISC-V and systems outside the usual x86 path.

2026.10 ACTIVE

RISC-V as a real system

RISC-V / Linux / SBC / Toolchains
2026.10 ACTIVE

From an amd64 pipeline to multi-architecture CI

CRUX / CI / ARM / RISC-V / Packaging / Gitea
03

Infrastructure

Self-hosted systems and the infrastructure behind them.

2026.09 ACTIVE

LLM inference on a legacy GPU

llama.cpp / CUDA / Pascal / LLM
2026.10 ACTIVE

Local image generation pipeline

Image Generation / Prompt Engineering / Automation / Local Inference
2026.10 ACTIVE

Clean environments for building ports

CRUX / Packaging / Build Environments / Reproducibility
2026.10 ACTIVE

From an amd64 pipeline to multi-architecture CI

CRUX / CI / ARM / RISC-V / Packaging / Gitea
04

Experiments

Ideas, prototypes and tools that are still finding their shape.

2026.10 ACTIVE

Image generation on Apple Silicon

Image Generation / Apple Silicon / Local Inference / Benchmarking
2026.10 ACTIVE

Clean environments for building ports

CRUX / Packaging / Build Environments / Reproducibility