Relevance Labs - Windows GPU Data Science Workstation AMI

## Pricing

This AMI carries no additional software charge. You pay only for the underlying AWS infrastructure costs (EC2 instances, EBS storage, and data transfer). There is no hourly software fee on top of your chosen instance type.

## Why This Workstation

• **Windows-native GPU stack**: The only Marketplace AMI combining a full Windows desktop, NVIDIA GPU acceleration, and both R and Python data science ecosystems in a single image with no software fee.
• **Zero-configuration remote desktop**: Amazon DCV provides encrypted, high-performance remote access with no additional licensing - connect and start working immediately.
• **Unified development environment**: Switch seamlessly between RStudio for statistical analysis, JupyterLab for interactive notebooks, and Visual Studio for application development without managing multiple servers.

## Key Features and Buyer Outcomes

**Accelerate Deep Learning Without Driver Headaches**

Pre-installed NVIDIA drivers, CUDA Toolkit, and cuDNN eliminate the most common pain point in GPU computing setup. PyTorch and TensorFlow are pre-configured to leverage GPU acceleration out of the box.

**Complete Development Stack**

• Visual Studio 2022 and VS Code for application development
• PyCharm Community Edition for Python workflows
• RStudio for R-based statistical analysis
• JupyterLab and Jupyter Notebook for interactive exploration
• Python 3.x and R pre-installed with essential packages

**Production-Ready Data Science Frameworks**

• PyTorch and TensorFlow for deep learning
• scikit-learn for classical machine learning
• PySpark and Dask for distributed data processing
• Vowpal Wabbit for online learning
• Anaconda for reproducible environment management

**Containerization and DevOps**

Docker and Docker Compose are pre-installed, enabling containerized workflows and reproducible experiments directly on the workstation.

**Productivity Tools**

Google Chrome, Git, AWS CLI, 7-Zip, and LibreOffice (Writer, Calc, Impress) are included for day-to-day productivity.

## Example Use Case

A financial analyst needs to train gradient-boosted models on transaction data using PySpark for preprocessing and scikit-learn for modeling. They launch this AMI on a GPU-enabled instance, connect via Amazon DCV, open JupyterLab, and begin processing data immediately - skipping the typical multi-hour setup of drivers, frameworks, and IDE configuration.

## AWS Integration

Designed to work alongside AWS services including Amazon SageMaker for managed training jobs, Amazon Bedrock for generative AI, and Amazon OpenSearch for search and analytics workloads. The pre-installed AWS CLI enables direct interaction with S3, EC2, and other services.

## Recommended Instance Types

• **GPU workloads (deep learning, model training)**: g4dn.xlarge, g5.xlarge, p3.2xlarge
• **CPU-only workloads (data analysis, R development)**: m5.xlarge, m5.2xlarge
• **Minimum recommended**: 4 vCPUs, 16 GB RAM, 100 GB EBS storage

## Getting Started

• Launch the AMI from AWS Marketplace on a supported instance type
• Configure your security group to allow TCP port 8443 (Amazon DCV)
• Connect via Amazon DCV client using your instance credentials
• Begin working in JupyterLab, RStudio, or your preferred IDE

Need help validating the workstation for your use case? Contact Relevance Labs at rlcloudsupport@relevancelab.com to schedule a guided setup walkthrough.

## Security Considerations

Amazon DCV sessions are encrypted in transit. Buyers should configure security groups to restrict port 8443 access to trusted IP ranges and follow AWS best practices for key pair management and IAM role assignment.

## Technical Details

• **Operating System**: Windows Server 2022
• **Remote Access**: Amazon NICE DCV (encrypted)
• **GPU Support**: NVIDIA drivers, CUDA Toolkit, cuDNN
• **Languages**: Python 3.x, R
• **IDEs**: Visual Studio 2022, VS Code, PyCharm CE, RStudio
• **Notebooks**: JupyterLab, Jupyter Notebook
• **Frameworks**: PyTorch, TensorFlow, scikit-learn, PySpark, Dask, Vowpal Wabbit
• **Environment Management**: Anaconda, Docker, Docker Compose