MLops Into Raspberry Pi 5

  • 23/2/2024
  • 4-minute read

One of the tools I use most for practicing MLOps, both for designing pipelines and APIs (for inference), is the Raspberry Pi. Today, I spent several hours trying to install Visual Studio Code to complement my iPad Pro as a development tool.

Why this setup? 🤔

  • Improve programming skills—I am a big fan of using Weights & Biases (W&B) to monitor the resource usage of each service I create.
  • Using the Raspberry Pi as a server allows me to test Edge computing deployments.
  • For scalable prototype development, it’s a great way to test artifacts and the lifecycle of models.
  • When designing a model from hyperparameters, it helps me fine-tune grid search or Bayesian methods efficiently to optimize experimentation.
  • Running MLflow on Edge computing enables optimization in model registry and updates.
  • Using Docker and Kubernetes helps ensure clean code before committing changes.

There are many more reasons, but these are the main ones. Now, how do you set up Raspberry Pi to unlock its full power for MLOps?

🔧 Setting Up Raspberry Pi for MLOps

First, install the Raspberry Pi OS. There are many tutorials, but I prefer the official documentation:

🔗 Raspberry Pi OS Installation

Next, find the Raspberry Pi’s IP address to connect to it from the iPad or another computer:

hostname -I

This will return something like 192.168.1.100 2601:123456. You can then connect via SSH:

ssh pi@192.168.1.100

After entering the password (set during installation), welcome to your new server! 🎉

1️⃣ Installing Conda on Raspberry Pi

By default, Python comes pre-installed. Now, install Conda using:

wget https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-aarch64.sh
bash Miniforge3-Linux-aarch64.sh

2️⃣ Installing MLflow

Once Conda is installed, install MLflow:

pip install mlflow

3️⃣ Installing Docker

To set up Docker, use:

sudo apt update && sudo apt upgrade -y
sudo apt install -y docker.io

For advanced edge computing, you can explore DockerPi: 🔗 DockerPi

4️⃣ Installing Tmux and Neovim

To improve terminal workflow and coding experience, install Tmux and Neovim:

sudo apt update && sudo apt install -y tmux

For Neovim:

sudo apt update
sudo apt install -y neovim python3-neovim

🔹 Neovim Configuration for Python Development

To configure Neovim for a better coding experience:

mkdir -p ~/.config/nvim
nano ~/.config/nvim/init.vim

Paste the following Neovim config for line numbers, syntax highlighting, and autocompletion:

" --- General Settings ---
set number            " Show line numbers
set mouse=a           " Enable mouse support
set cursorline        " Highlight the current line
set expandtab         " Use spaces instead of tabs
set shiftwidth=4      " Indentation size
set tabstop=4         " Tab size
set autoindent        " Maintain indentation
set smartindent       " Smart indentation
set background=dark   " Dark theme
set termguicolors     " Enable true colors
set encoding=utf-8    " UTF-8 support
syntax on             " Enable syntax highlighting
filetype plugin indent on " Enable plugins and indentation

" --- Plugins ---
call plug#begin('~/.vim/plugged')

" File explorer
Plug 'preservim/nerdtree'

" Autocompletion
Plug 'neoclide/coc.nvim', {'branch': 'release'}

" Language server support
Plug 'neovim/nvim-lspconfig'

" Improved syntax highlighting
Plug 'nvim-treesitter/nvim-treesitter', {'do': ':TSUpdate'}

" Status bar
Plug 'vim-airline/vim-airline'

call plug#end()

" --- Keybindings ---
nnoremap <C-n> :NERDTreeToggle<CR> " Open/close file explorer
nnoremap <C-p> :Files<CR> " Quick file search

" --- coc.nvim Configuration ---
let g:coc_global_extensions = ['coc-python', 'coc-json', 'coc-html', 'coc-tsserver']

" Auto-format Python code on save
autocmd BufWritePre *.py :Black

5️⃣ Installing and Configuring VS Code (code-server)

🚀 Installing code-server

curl -fsSL https://code-server.dev/install.sh | sh

🚀 Enabling code-server

sudo systemctl enable --now code-server@$USER

If the service is masked:

sudo systemctl unmask code-server@$USER
sudo systemctl restart code-server@$USER

🚀 Checking code-server status

sudo systemctl status code-server@$USER

If active (running) appears, code-server is working.

6️⃣ Configuring code-server for iPad Access

Edit the config file:

nano ~/.config/code-server/config.yaml

Set the following configuration:

bind-addr: 0.0.0.0:8080  # Change port if needed
auth: none               # Disable authentication (or use password)
cert: false              # No HTTPS

Save (Ctrl + X, Y, Enter), then restart:

sudo systemctl restart code-server@$USER

Check the correct port:

sudo netstat -tulpn | grep LISTEN

7️⃣ Accessing VS Code from the iPad

Open Safari or Chrome on the iPad and enter:

http://<RASPBERRY_PI_IP>:8080

For example:

http://192.179.1.100:8080

To find your Raspberry Pi’s IP:

hostname -I

✅ If everything is correct, VS Code (code-server) will open in the browser.

8️⃣ Troubleshooting

🔹 If code-server doesn’t load

  1. Check if code-server is running: 
    sudo systemctl status code-server@$USER
  2. Try accessing from Raspberry Pi itself: 
    curl -v http://127.0.0.1:8080
  3. Ensure Raspberry Pi and iPad are on the same network.

🔹 If the port is occupied (EADDRINUSE)

  1. Kill previous processes: 
    pkill -f code-server
  2. Restart code-server: 
    sudo systemctl restart code-server@$USER
  3. Verify the port: 
    sudo netstat -tulpn | grep LISTEN

With this setup, your Raspberry Pi 5 becomes a powerful MLOps workstation. 🚀🔥
Let me know in the comments if you have questions!

Edge ComputingMLOps raspberry-pimlflowedge-ai