https://github.com/brevdev/setup-scripts

Brev Setup Scripts
https://github.com/brevdev/setup-scripts

Last synced: 6 months ago
JSON representation

Brev Setup Scripts

• Host: GitHub
• URL: https://github.com/brevdev/setup-scripts
• Owner: brevdev
• License: apache-2.0
• Created: 2022-04-05T03:30:20.000Z (about 4 years ago)
• Default Branch: main
• Last Pushed: 2025-10-18T05:45:55.000Z (6 months ago)
• Last Synced: 2025-10-18T21:42:05.640Z (6 months ago)
• Language: Shell
• Homepage: https://developer.nvidia.com/brev
• Size: 106 KB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Brev Setup Scripts

Simple, practical setup scripts for common developer environments.

**What Brev already provides:** NVIDIA drivers, CUDA toolkit, Docker, NVIDIA Container Toolkit

## Available Scripts

### 🐍 Python Development

```bash

cd python-dev && bash setup.sh

```

**Installs:** pyenv, Python 3.11, Jupyter Lab, common packages (pandas, numpy, matplotlib)  

**Time:** ~3-5 minutes

### 📦 Node.js Development

```bash

cd nodejs-dev && bash setup.sh

```

**Installs:** nvm, Node LTS, pnpm, TypeScript, ESLint, Prettier  

**Time:** ~2-3 minutes

### 💻 Terminal Setup

```bash

cd terminal-setup && bash setup.sh

```

**Installs:** zsh, oh-my-zsh, fzf, ripgrep, bat, eza (modern CLI tools)  

**Time:** ~2-3 minutes  

**Note:** Automatically switches to zsh when complete

### ☸️ Local Kubernetes

```bash

cd k8s-local && bash setup.sh

```

**Installs:** microk8s, kubectl, helm, k9s, GPU operator  

**Time:** ~3-5 minutes  

**Note:** All tools work immediately, no group membership or logout needed

### 🤖 ML Quickstart

```bash

cd ml-quickstart && bash setup.sh

```

**Installs:** Miniconda, PyTorch with CUDA, Jupyter Lab, transformers  

**Time:** ~5-8 minutes (PyTorch is large)

### ⚡ RAPIDS

```bash

cd rapids && bash setup.sh

```

**Installs:** GPU-accelerated pandas (cuDF), scikit-learn (cuML), example notebooks  

**Time:** ~8-12 minutes  

**Note:** 10-50x faster data processing on GPU. Requires NVIDIA GPU

### 🦙 Ollama

```bash

cd ollama && bash setup.sh

```

**Installs:** Ollama with GPU support, llama3.2 model (pre-downloaded)  

**Time:** ~3-5 minutes  

**Port:** 11434/tcp for API access

### 🚀 Unsloth

```bash

cd unsloth && bash setup.sh

```

**Installs:** Unsloth for fast fine-tuning, PyTorch with CUDA, LoRA/QLoRA support  

**Time:** ~5-8 minutes  

**Note:** Requires NVIDIA GPU

### 🔄 LiteLLM

```bash

cd litellm && bash setup.sh

```

**Installs:** Universal LLM proxy (use any LLM with OpenAI API format)  

**Time:** ~1-2 minutes  

**Port:** 4000/tcp for API access

### 🔍 Qdrant

```bash

cd qdrant && bash setup.sh

```

**Installs:** Vector database for RAG and semantic search  

**Time:** ~1-2 minutes  

**Port:** 6333/tcp for API + dashboard

### 🎨 ComfyUI

```bash

cd comfyui && bash setup.sh

```

**Installs:** Node-based UI for Stable Diffusion, SD 1.5 model  

**Time:** ~5-10 minutes  

**Port:** 8188/tcp for web interface  

**Note:** Requires NVIDIA GPU

### 🗄️ Databases

```bash

cd databases && bash setup.sh

```

**Installs:** PostgreSQL 16, Redis 7 (in Docker containers)  

**Time:** ~1-2 minutes

### 📓 Marimo

```bash

cd marimo && bash setup.sh

```

**Installs:** Marimo reactive notebooks as systemd service  

**Time:** ~2-3 minutes  

**Port:** 8080/tcp for web access

### 🛡️ earlyoom

```bash

cd earlyoom && bash setup.sh

```

**Installs:** Early OOM daemon to prevent system freezes  

**Time:** ~1-2 minutes  

**Note:** Monitors memory/swap and kills processes before OOM hangs

## Quick Start

**Pick what you need:**

```bash

# Python ML developer

cd ml-quickstart && bash setup.sh

# Web developer

cd nodejs-dev && bash setup.sh

cd databases && bash setup.sh

# Terminal power user

cd terminal-setup && bash setup.sh

# Kubernetes developer

cd k8s-local && bash setup.sh

```

## Design Philosophy

Each script is:

- ✅ **Simple** - One purpose, no complexity

- ✅ **Short** - Under 150 lines each

- ✅ **Fast** - Takes 2-8 minutes

- ✅ **Standalone** - No dependencies between scripts

- ✅ **Practical** - Installs what developers actually use

We don't:

- ❌ Install what Brev already provides (NVIDIA drivers, CUDA, Docker)

- ❌ Add complex GPU detection logic

- ❌ Support multi-node/HPC scenarios

- ❌ Over-engineer solutions

## Examples

**Python data science:**

```bash

cd python-dev && bash setup.sh

# Then:

ipython

jupyter lab --ip=0.0.0.0

```

**Machine learning with GPU:**

```bash

cd ml-quickstart && bash setup.sh

# Then:

conda activate ml

python gpu_check.py

```

**GPU-accelerated data science with RAPIDS:**

```bash

cd rapids && bash setup.sh

# Then:

conda activate rapids

python ~/rapids-examples/benchmark.py  # See 20x+ speedup!

```

**Local LLM with Ollama:**

```bash

cd ollama && bash setup.sh

# Then:

ollama run llama3.2

ollama list

```

**Fast LLM fine-tuning with Unsloth:**

```bash

cd unsloth && bash setup.sh

# Then:

conda activate unsloth

python ~/unsloth-examples/test_install.py

```

**Universal LLM proxy with LiteLLM:**

```bash

cd litellm && bash setup.sh

# Then use any LLM with OpenAI SDK:

# openai.api_base = "http://localhost:4000"

```

**Vector database with Qdrant:**

```bash

cd qdrant && bash setup.sh

# Then:

pip install qdrant-client

python ~/qdrant_example.py

```

**Image generation with ComfyUI:**

```bash

cd comfyui && bash setup.sh

# Then open: http://localhost:8188

```

**Modern terminal:**

```bash

cd terminal-setup && bash setup.sh

# Automatically drops you into zsh, then:

ll    # Better ls

cat file.txt  # Syntax highlighting

fzf   # Fuzzy finder

```

**Local database:**

```bash

cd databases && bash setup.sh

# Then:

docker exec -it postgres psql -U postgres

docker exec -it redis redis-cli

```

**OOM protection with earlyoom:**

```bash

cd earlyoom && bash setup.sh

# Then:

sudo systemctl status earlyoom

sudo journalctl -u earlyoom -f  # Watch memory monitoring

```

## File Structure

```

brev-setup-scripts/

├── README.md                    # This file

├── python-dev/

│   ├── setup.sh                 # Python development environment

│   └── README.md

├── nodejs-dev/

│   ├── setup.sh                 # Node.js development environment

│   └── README.md

├── terminal-setup/

│   ├── setup.sh                 # Modern terminal with zsh

│   └── README.md

├── k8s-local/

│   ├── setup.sh                 # Local Kubernetes

│   └── README.md

├── ml-quickstart/

│   ├── setup.sh                 # PyTorch ML environment

│   └── README.md

├── ollama/

│   ├── setup.sh                 # Ollama LLM inference

│   └── README.md

├── unsloth/

│   ├── setup.sh                 # Unsloth fast fine-tuning

│   └── README.md

├── litellm/

│   ├── setup.sh                 # Universal LLM proxy

│   └── README.md

├── qdrant/

│   ├── setup.sh                 # Vector database

│   └── README.md

├── comfyui/

│   ├── setup.sh                 # ComfyUI for Stable Diffusion

│   └── README.md

├── databases/

│   ├── setup.sh                 # PostgreSQL + Redis

│   └── README.md

├── marimo/

│   ├── setup.sh                 # Marimo reactive notebooks

│   └── README.md

├── earlyoom/

│   ├── setup.sh                 # Early OOM daemon

│   └── README.md

└── rapids/

    ├── setup.sh                 # RAPIDS GPU-accelerated data science

    └── README.md

```

## Contributing

Want to add a script? Keep it simple:

1. **One purpose** - Install one thing well

2. **Short** - Under 150 lines

3. **Fast** - Completes in < 10 minutes

4. **Verify** - Include a verification step

5. **Document** - Show quick start commands

## License

Apache 2.0