Customer-Hosted AI
For organizations with existing data center infrastructure, EQ Syntropy can be deployed on customer servers with full AI capabilities.
Overview
Customer-hosted deployment is designed for:
- Enterprise environments with established IT infrastructure
- Organizations with strict data residency requirements
- Air-gapped or restricted network environments
- Multi-site deployments requiring centralized management
AI inference runs locally with no per-query fees, no usage limits, and no external dependencies.
Requirements
Minimum Server Specifications
- CPU: 8+ cores, x86-64 architecture
- RAM: 32GB minimum, 64GB recommended
- Storage: 1TB NVMe SSD for application and recent data
- Network: Gigabit Ethernet connectivity to gateway devices
Recommended GPU (for full AI capabilities)
- NVIDIA GPU: RTX 4090 or A6000 class
- VRAM: 24GB minimum
- Driver: CUDA 12.x compatible
Operating System
- Debian 12 (Bookworm) or Ubuntu 22.04 LTS
- Docker and Docker Compose for containerized deployment
- Systemd for service management
Architecture
Customer-hosted deployments typically include:
- Application Server: EQ Syntropy core services (Rust backend, Python tools)
- Data Storage: Structured parquet files for waveform data, DuckDB for indexed queries
- GPU Server: Optional, for AI inference workloads
Installation
Customer-hosted deployments use the same gateway software package as flex gateway deployments, installed on customer infrastructure.
Contact [email protected] for installation guidance and enterprise deployment assistance.