# STELS - Runtime Engineering

## **Introduction**

<figure><img src="https://4190256927-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F1P2s3me8iEqQ7X2aWrKW%2Fuploads%2FwrZafWJkyL7tvvrG2LsL%2FDesign.svg?alt=media&#x26;token=c361d5e7-4053-4403-8111-924671000dfc" alt="" width="394"><figcaption></figcaption></figure>

```sh
STELS: This [DRAFT] version 0.29
```

The **STELS** platform is a Decentralized System Designed for Automating Digital Markets.&#x20;

It provides tools for programming and training intelligent trading models. It supports both professional market participants and beginner traders, known as **Cyber Traders** in the STELS ecosystem.&#x20;

The platform has been developed by leading experts in neural networks and automation, adhering to modern regulatory requirements.

**Key Features of STELS:**

* [x] Programming and training intelligent trading models.
* [x] Configuring Neural Networks for Collective Intelligence.
* [x] Support for both programmers and professionals in automation and Machine Learning.
* [x] Hybrid integration with cryptocurrency exchanges.
* [x] Support for up to 5000 exchange accounts (in the AMI Nominator version).
* [x] Creation and customization of private/public pools.
* [x] Marketplace for Trained Models and Dashboard Visualization Widgets.
* [x] Dynamic code generation for strategies.
* [x] Heterogeneous multi-layered private/public node network.

**System Requirements**

To effectively run the STELS platform, your system should meet the following requirements:

1. **Minimum 4 GB RAM** for basic operations.
2. **Minimum 4-core processor** for handling multiple threads efficiently.
3. **Operating System:** Linux, macOS, or Windows.
4. **High-speed internet connection** to ensure seamless data transfers and interaction with exchanges.
5. **Support for CPU, GPU, and NPU:**
   * **CPU:** Required for core processing.
   * **GPU:** Significantly accelerates model training and decision-making processes.
   * **NPU:** Currently in experimental stages, benchmarks will be released soon.
6. **SSD storage** for fast data access and improved performance.
7. **At least 600 GB bandwidth** for network-intensive operations, especially when interacting with multiple exchanges and handling large amounts of data.

> **Note:** The more computational resources available, the faster data processing, neural network training (RNN), and the runtime execution of trading operations will be. Utilizing **GPUs** can substantially boost training speed and decision-making, while **NPU** support is under testing and will be optimized based on upcoming benchmarks.