Improve Jobs docs onboarding and Quickstart

[davanstrien]

Starting a focused refresh of the Jobs docs to make it clearer what Jobs is for, how to run a first workload, and where to go next.

This first pass:

• Clarifies the landing and overview introductions and adds a clearer route into the Quickstart.
• Reshapes the Quickstart around a CPU Hello World followed by a small GPU inference example.
• Includes the example script in this repo so readers can inspect it and run it directly by URL.
• Points readers toward training, inference and data-processing examples.

---

Note

Low Risk
Documentation and a sample script only; no product code, auth, or runtime behavior changes.

Overview
Refreshes Hugging Face Jobs onboarding docs so new readers get a clearer value proposition and a shorter path to a first successful run.

The Jobs landing page (jobs.md), overview (jobs-overview.md), and feature grid copy are rewritten with direct links into the Quickstart and examples. Long introductory paragraphs are trimmed, and the “run any workload” section is split into clearer Python (hf jobs uv run) vs Docker (hf jobs run) subsections with updated grid labels (e.g. hardware and “run your code” wording).

The Quickstart is rebuilt around a CPU Hello World, then a small GPU text-generation example on t4-small—replacing the previous TRL fine-tuning walkthrough, extra CLI install variants, Docker hello-world, and Hub UI screenshots. It documents hf jobs inspect / logs / cancel, optional local script upload, and a Next steps list pointing to training, datasets, Docker, and agent examples.

Adds examples/jobs/hello_gpu.py in-repo so the guide can run the sample via a raw GitHub URL or a local file.

^{Reviewed by Cursor Bugbot for commit 4467781. Bugbot is set up for automated code reviews on this repo. Configure here.}

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[davanstrien]

cc @lhoestq @julien-c @gary149 @Wauplin for viz. Will add a few more changes to this PR but try and keep the scope small enough it's easy to review still.

[davanstrien]

First pass on a bigger Jobs docs refresh. I don’t want to put everything into one PR, but for context I also planning to work on:

• clearer guidance on using Docker images, including images built from Docker Spaces.
• A longer inference/data-processing walkthrough, covering bringing data into a Job, saving results with Buckets, and adapting the example to your own workload.
• not quite sure exactly how I will do it yet, but a bit of a "how to do X in Jobs" i.e. how mounting volumes can work, how you can use ports etc.

This PR focuses on making it easier to understand Jobs and get a first successful run.