Get Started

Privacy Policy

empirischtech Inference VM — Azure Marketplace

Last updated: May 2026

1. Introduction

This privacy policy applies to the empirischtech Inference VM ("the Product") published on the Microsoft Azure Marketplace by Empirisch Tech GmbH ("we", "us", "our"). The Product is a pre-configured Ubuntu Linux virtual machine optimized for AI inference workloads, featuring NVIDIA GPU support, Docker containerization, and the vLLM inference engine.

2. Responsible Party

Empirisch Tech GmbH
Vienna, Austria
Email: contact@empirischtech.at
Phone: +43-1-2396082

3. Data We Collect

We do not collect, store, process, or transmit any personal data or usage data from customers who deploy or use this virtual machine image.

The Product is a self-contained operating system image. Once deployed to your Azure subscription:

  • The virtual machine runs entirely within your own Azure tenant and subscription.
  • We have no access to the virtual machine, its file system, network traffic, inference requests, or any data stored on it.
  • We do not embed any telemetry, analytics, tracking software, or phone-home mechanisms in the image.
  • We do not collect IP addresses, usage statistics, login information, model inputs/outputs, or any other data from deployed instances.

4. External Data Connections

First Boot Model Download: On initial startup, the VM downloads a pre-trained AI model from Hugging Face (specifically: empirischtech/DeepSeek-R1-Distill-Qwen-32B-gptq-4bit). This download:

System Updates: The Product includes standard Ubuntu packages that may connect to Ubuntu update servers when you initiate system updates. These connections are governed by Canonical's privacy policy.

Your Use of Microsoft Azure: Your Azure subscription and usage are governed by Microsoft's privacy statement. Microsoft may collect usage and billing data as described in their policies.

5. Cookies & Tracking

The Product itself does not use cookies or any form of user tracking. If you are viewing this page on our website (empirischtech.at), please refer to our general website privacy policy for information about cookies used on this website.

6. Data Storage & Security

Since we do not collect any data from the Product, there is no personal data stored on our systems related to your use of this virtual machine image. All data created during your use of the Product — including AI model inputs, outputs, logs, and any files you upload — resides exclusively within your Azure environment and is subject to your own data management and security policies.

7. Your Rights (GDPR)

The General Data Protection Regulation (GDPR) grants you rights such as access, rectification, erasure, restriction of processing, data portability, and objection regarding your personal data. Since this Product does not collect, store, or process any personal data whatsoever, there is no personal data held by us that could be returned, modified, or deleted. Consequently, these rights do not apply in the context of this Product.

8. Changes to This Policy

We may update this privacy policy from time to time. Any changes will be posted on this page with an updated revision date. We encourage you to review this policy periodically.

9. Contact

If you have any questions about this privacy policy, our data practices, or the empirischtech Inference VM, please contact us:

Empirisch Tech GmbH
Vienna, Austria
Email: contact@empirischtech.at
Phone: +43-1-2396082
Website: empirischtech.at

10. Software Licenses

The empirischtech Inference VM bundles multiple open-source and proprietary software components. For complete license information, including third-party software notices, please refer to the LICENSE.md file included in the VM at /opt/empirischtech/LICENSE.md or the canonical web notice: Third-party software licenses.

Key components include:

  • Ubuntu 24.04 LTS (GPL, LGPL, BSD, MIT per-package)
  • NVIDIA Drivers & CUDA (NVIDIA Software License Agreement)
  • Docker CE, Caddy (Apache-2.0)
  • vLLM, PyTorch, Hugging Face libraries (Apache-2.0, BSD-3-Clause)
  • Pre-trained model (CC-BY-4.0, downloaded from Hugging Face)