Northflank is the developer platform you’ll use to deploy your work, using compute provided by Coreweave.

tl;dr

• To get started, you first need to request access to the compute by filling out this form.
• Every team gets an H100.
• You will receive an email from Northflank from [email protected].
• Accept your Northflank Team invite in your email, this will open your Northflank Team dashboard
• In your Northflank Project called hackathon, you can create a Northflank Service (container) with access to your H100. You can follow our guides to deploy :
  • Recommended: a Jupyter notebook w/ PyTorch preinstalled –Jupyter Notebook with PyTorch
  • A generic Ubuntu image – Shell access (VM-like experience)
  • Your existing GitHub repository using Docker – CI/CD Python application
  • Or any Public Docker image – Common AI Docker Images
• You can then follow our guide for Using SSH to access your Services to connect to your instance w/ VSCode/Cursor/etc
• In case you need to troubleshoot something, reference the Use and troubleshoot your deployments or mention anyone from the Northflank team in the OpenEnv Discord channel

In-depth guide

1. Your team will have a Northflank team and project associated to you. You’ll receive an invite for your team to your email address:

1. Once you’ve made an account and accepted the invite, you’ll be presented with the Northflank home page. Here you can find your project called Hackathon. You can click on it and this will navigate you to the project dashboard.

1. Your team will have access to a GPU node with a single NVIDIA H100 for your task.
2. You can then deploy your code to Northflank, for example
   • Recommended: a Jupyter notebook w/ PyTorch preinstalled –Jupyter Notebook with PyTorch
   • A generic Ubuntu image – Shell access (VM-like experience)
   • Your existing GitHub repository using Docker – CI/CD Python application
   • Or any Public Docker image – Common AI Docker Images

<aside> 💡

You can only have 1 GPU service, if you need more, talk with the organizers to see if there’s more capacity available. You can scale down your existing service if you want to try something else.

</aside>

Extra information

• The development sandboxes are running Ubuntu 22.04.5 and are based off the pytorch/pytorch:2.8.0-cuda12.6-cudnn9-runtime Docker image. Any system packages you install or changes you make to your environment won’t be persisted upon reboot.
• You can install the Northflank CLI on your machine (https://northflank.com/docs/v1/api/use-the-cli) and easily upload/download files to/from the sandbox, e.g. to visualise ncu/nsys reports using the NVIDIA GUI tools
  • You can install the CLI on your laptop using npm or yarn - yarn global add @northflank/cli npm i -g @northflank/cli
  • After you’ve installed it, you can run northflank login and follow the instructions to authenticate your CLI with your Northflank team.