CPP DS&AI Kaggle Team — CloudDock Sponsorship

Why CloudDock for Kaggle?

Single-tenant GUI desktop: your own dedicated machine. No noisy neighbors.
Transparent queue: see your position, ETA, and the exact end time before starting.
Predictable runtime: you choose the hours; sessions won’t be preempted mid-run.
One-click Jupyter: launches in seconds; use the full desktop for browsers, editors, viewers.
Pre-installed ML stack: common packages are ready; otherwise just pip install.
CloudDock App Store: one-click install of popular apps (VS Code, LibreOffice, VLC, etc.).
Stability & safety: CloudDock Intelligent Dog™ monitor containers for health and abuse.

Why CloudDock is different

Unlike Colab, CloudDock never cuts your session short because you’re not paying.
Your booked hours are guaranteed — stable, predictable, transparent.
But CloudDock is not a school computer lab. This is the real world: resources are finite, rules apply, everyone shares fairly.

Sponsorship means opportunity, not exemption. Welcome to CloudDock.

CloudDock vs Colab (at a glance)

Feature	CloudDock	Colab (typical)
Runtime predictability	Fixed by you; visible end time	Session caps; may disconnect
Queue transparency	Shows position & ETA	Not visible
Tenancy	Single-tenant GUI desktop	Shared
Jupyter startup	One-click, seconds	Varies
Full desktop apps	Yes (Firefox, App Store)	No (notebook-only)
Environment hassle	Curated, ready to go	Depends on runtime
Auto cleanup	Container auto-destroy at end time	N/A

How to get sponsored (what to do)

CPP (via Discord)

1 Contact us on Discord → chat.

2 Tell us how many accounts you need and the proposed usernames. one team group → one account

3 We review & approve and confirm in Discord.

4 Members register those usernames on CloudDock.

5 DM us the final usernames → we link them and credits start weekly.

Other schools (via Telegram)

1 Contact customer support on Telegram → chat.

2 Provide your application ID, school email, club name, and student ID.

3 Tell us how many accounts you need (one team group → one account).

4 We review & approve and confirm in Telegram.

5 Members register → send us the registered usernames → we link and start weekly credits.

Sessions end at the booked time; capacity is shared via a transparent queue. Sponsorship covers usage credits only.

Eligibility & scope

Who: active CPP DS&AI Kaggle Team members on the officer-verified roster.
Use cases: Kaggle competitions, team projects, model exploration, notebooks/report pipelines.
Period: 12 months; renewals considered with continued activity and feedback.
Deliverables: brief monthly feedback + bug reports (stability, performance, UX).

Credits & usage policy

Weekly credits
Sponsored accounts receive a recurring weekly credit top-up. Unused credit may not roll over (see Terms).

No special privileges
Same availability rules for everyone. If machines are busy, you’ll queue. No reserved capacity.

Fair use
Plan hours, end on time, keep workloads reasonable. Abuse triggers CloudDock Intelligent Dog™ policies.

Sponsorship covers usage credits only; it does not grant priority scheduling.
Teams manage their own time within quota. CloudDock doesn’t provide tutoring or coursework help.
All users (sponsored or paid) share the same transparent queueing system to ensure fairness.

Redeeming your credits

Log in → Dashboard → choose a group (e.g., A/B/C) → set hours until your credit covers it.
Confirm: you’ll see the exact end time before launch.
Run Jupyter (one-click) or use the full desktop.

What’s pre-installed?

Current Kaggle image (CUDA 12.1):

Desktop & UX: Firefox (non-snap), tuned XFCE (Arc-Lighter, Papirus-Light, Ubuntu fonts).
PyTorch: torch==2.3.1+cu121, torchvision==0.18.1+cu121, torchaudio==2.3.1+cu121.
HF/Training: diffusers, transformers, accelerate, safetensors, einops, datasets, evaluate, tensorboard, tqdm.
Sci/Data: pandas, scikit-learn, sympy, librosa==0.10.1, numba==0.59.1, llvmlite==0.42.0.
Vision/Media: opencv-python-headless==4.8.1.78, pillow, imageio, imageio-ffmpeg, moviepy.
ONNX: onnx==1.16.0, onnxruntime==1.17.1, with numpy<2.0, protobuf<5.0.
Web/Service: pyarrow==16.1.0, gradio==4.44.1, streamlit==1.36.0, fastapi==0.115.6, uvicorn==0.31.0.
Lab/Monitor: jupyterlab==4.2.5, notebook==7.2.1, jupyter_server<3, jupyterlab_server<3, gpustat==1.1.1, nvidia-ml-py3==7.352.0, psutil==5.9.8, blessed==1.20.0.
Lightning: pytorch-lightning==2.2.4; xFormers: 0.0.27.post2 → 0.0.25.post1 fallback.

Common extras you might add

pip install kaggle seaborn matplotlib lightgbm xgboost catboost polars

Quick Start

Login → Dashboard → choose a machine group → set hours → Start.
Click Jupyter (one-click) or open the GUI desktop from the dashboard.
Use the in-container browser to sign in to Google Drive/iCloud/etc. and download to ~/Downloads.

Pick hours; your exact end time is shown before launch.

One-click Jupyter; or use the full XFCE desktop.

Transparent Queue (position & ETA)

You can see your queue position, expected start, and exact end time before launching. It’s easy to plan runs and presentations without surprise interrupts.

Transparent queue view with position and ETA

Know your place in line and when the session will end.

Full GUI Desktop

A full XFCE desktop runs in your browser — use a file manager, web browser, terminal, and the CloudDock App Store. No SSH needed. Great for demos, grading, and quick debugging.

Open Downloads to find files from the in-container browser.
Launch tools from the App Store (VS Code, LibreOffice, VLC, etc.).
Combine with one-click Jupyter for a seamless workflow.

XFCE desktop with browser and Downloads inside the container

A full desktop in the browser — files, browser, editors, Jupyter.

Kaggle API (optional)

# 1) Upload kaggle.json via the GUI (Downloads), then:
mkdir -p ~/.kaggle && mv ~/Downloads/kaggle.json ~/.kaggle/ && chmod 600 ~/.kaggle/kaggle.json

# 2) Download data/competitions
kaggle datasets download -d zynicide/wine-reviews -p ~/data
kaggle competitions download -c titanic -p ~/data

Security & Reliability

CloudDock Intelligent Dog™: monitor processes, block abusive patterns, preserve incident logs.
Auto-destroy lifecycle: when your booked time ends, the container is securely destroyed.
Graceful shutdown: sessions end at the booked time; the platform shuts down cleanly.

Stability first: supervised containers, predictable lifecycle.

Fair by design

No hidden limits, no “second-class” accounts.
Sponsored users and paying clients follow the same clear rules.
Your sessions won’t be cut off unfairly — we reject “unequal treaties.”
True fairness is the most thoughtful form of support.

No special privileges

Sponsored accounts follow the same availability rules as everyone else.
If all machines are busy, you’ll enter the transparent queue.
If a GPU group is fully booked, it shows Full until a slot frees up.
Sponsorship = usage credits, not priority nor reserved capacity.

FAQ

Will my session stop early?

No. Your booked hours are honored; the exact end time is shown.

Can I install extra packages or apps?

Yes. Use pip install ... for Python libs, and the CloudDock App Store for desktop apps.

Where are my downloads?

Use the in-container browser; files land in ~/Downloads. Move them into your project folder as needed.

“One team, one desktop. Train, iterate, submit.”

By using CloudDock you agree to our Terms of Service and Privacy Policy. Sponsorship rules may change; the latest Terms prevail.

← Back to Documents