Skip to content
View CastilloworksAi's full-sized avatar

Block or report CastilloworksAi

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. 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
CastilloworksAi/README.md

Hey, I'm Castillo

I build AI tools out of Texas, and almost all of it runs on my own machine. No cloud bills, no data leaving the house.

Most of what's here started as something I needed for my own work: image pipelines, LoRA training, voice cloning, a local chat app, and a pile of terminal tools I use every day.

The rig

Everything in these repos is built and tested on this box:

Part What I run
CPU AMD Ryzen 9 9900X (12 cores)
GPU NVIDIA RTX 5080, 16 GB (Blackwell)
RAM 96 GB DDR5-6000 (2x 48 GB)
Board Gigabyte X870E AORUS ELITE WIFI7
Storage 2x 4 TB Kingston Fury Renegade NVMe (Gen4), 2 TB WD external
OS Ubuntu 24.04 LTS, kernel 7.0
Driver NVIDIA 580.178 (CUDA 13.0)
PyTorch 2.11 + cu128

Software on top: ComfyUI for images, Ollama for local LLMs, Claude Code for building. Nothing AI starts at boot. I turn each service on when I need it.

Tools

Repo What it does
imagine Type a prompt in the terminal, get a PNG. A small CLI for ComfyUI
headshot Selfie in, studio headshots out, on your own GPU
clone Local voice cloning with Chatterbox. No per-character fees
grab Download video or audio from the web with one word
broll Terminal B-roll for screen recordings. Pure bash
terminal-fun Random terminal animations for recording content

Image models and training

Repo What it does
comfyui-workflows My ComfyUI workflows: Flux, SDXL, LoRA training, captioning
z-image-turbo-comfy One-command ComfyUI for Z-Image-Turbo on RTX 50-series cards
lora-training-guide Beginner guide, from raw images to a trained .safetensors
lora-studio Local web app for captioning datasets and exporting training jobs
ai-workstation Setup scripts, backups, and model downloaders for this rig

Other stuff

Repo What it does
block-buddies Place-value number game for kids. Moved to castilloworks.ai, repo archived
valet-trash-calculator Profit calculator for a valet trash business
killswitch Data wipe scripts for Linux, macOS, Windows, and ChromeOS. NUCLEAR tier is written and dry-runs by default; not yet run on live hardware

Stack

Python Shell ComfyUI Ollama CUDA Ubuntu

Based in Texas. More at castilloworks.ai.

Popular repositories Loading

  1. headshot headshot Public

    Turn a selfie into studio headshots on your own GPU. Generates the portrait in ComfyUI, then swaps your real face in with ReActor.

    Python 2

  2. comfyui-workflows comfyui-workflows Public

    My ComfyUI workflows: Flux, SDXL, LoRA training, captioning, and dataset building.

    1

  3. lora-studio lora-studio Public

    Local web UI for preparing Flux LoRA training datasets: captioning plus export to ComfyUI FluxTrainer. Single-file Python server, no cloud calls.

    Python 1

  4. ai-workstation ai-workstation Public

    Scripts and notes for a local Ubuntu 24.04 + RTX 5080 AI workstation: Blackwell-tuned ComfyUI launch, Ollama, nightly backups, model downloaders.

    Shell 1

  5. imagine imagine Public

    Type a prompt in the terminal, get a PNG. A small standard-library Python CLI for a local ComfyUI server.

    Python 1

  6. valet-trash-calculator valet-trash-calculator Public

    Python calculator for modeling valet trash business profit. Plug in your local numbers.

    Python