Architecture¶

The HCLS AI Factory runs three GPU-accelerated AI pipelines on a single NVIDIA DGX Spark workstation — from raw FASTQ sequencing data to ranked drug candidates in under 5 hours.

System Architecture¶

![HCLS AI Factory on NVIDIA DGX Spark — Architectural Infographic](diagrams/dgx_spark/HCLS%20AI%20Factory%20on%20NVIDIA%20DGX%20Spark%20Architectural%20Infographic%20v1.0.png){ loading=lazy }

Architectural Infographic (Alt View)¶

![HCLS AI Factory on NVIDIA DGX Spark — Architectural Infographic Alt View](diagrams/dgx_spark/HCLS%20AI%20Factory%20on%20NVIDIA%20DGX%20Spark%20Architectural%20Infographic%20v1.0%20%28Alt%20View%29.png){ loading=lazy }

Pipeline Logical Diagram¶

![HCLS AI Factory on NVIDIA DGX Spark — Pipeline Logical Diagram](diagrams/dgx_spark/HCLS%20AI%20Factory%20on%20NVIDIA%20DGX%20Spark%20Pipeline%20Logical%20Diagram%20v1.0.png){ loading=lazy }

draw.io Diagrams¶

High Level¶

![HCLS AI Factory — draw.io High Level](diagrams/dgx_spark/HCLS%20AI%20Factory%20on%20NVIDIA%20DGX%20Spark%20drawio%20Diagram-High%20Level%20v1.0.png){ loading=lazy }

Medium Level¶

![HCLS AI Factory — draw.io Medium Level](diagrams/dgx_spark/HCLS%20AI%20Factory%20on%20NVIDIA%20DGX%20Spark%20drawio%20Diagram-Medium%20Level%20v1.0.png){ loading=lazy }

Interactive Diagrams¶

Explore the architecture in detail with interactive draw.io diagrams:

Detailed Architecture — DGX Spark — Full component-level view
High-Level Architecture — DGX Spark — Simplified pipeline overview

Pipeline Mindmap¶

View Pipeline Mindmap (PDF)

Hardware Platform¶

Component	Specification
System	NVIDIA DGX Spark
GPU	GB10 Grace Blackwell Superchip
Memory	128 GB unified LPDDR5x
CPU	144 ARM64 cores (Grace)
Storage	NVMe SSD
Price	$3,999

Deep Dives¶

Project Bible — Complete technical reference with scoring formulas, thresholds, and configurations
White Paper — Architecture overview and design rationale
Architecture Mindmap — Visual system map