UP42 Storage: UP42 Processing Engine for Console

Integrate Processing Engine 2.0 into the UP42 Console, allowing users to apply advanced processing (e.g., car detection, vegetation index) directly to purchased imagery—without coding.

 
 

Why Build It?

Create a frictionless journey from data purchase → processing → insights.

The Challenge

Users needed to:

  1. Select assets (single or batch).

  2. Discover available processes (e.g., "Car Detection," "NDVI").

  3. Run and monitor jobs—while understanding backend implications (cost, runtime).

    Key Hurdle: Balancing simplicity with technical transparency.

Solution: A 3-Week Concept Sprint

1. Service Blueprinting (Behind the Scenes)

  • Mapped the full system flow:

    • User actions (e.g., selecting imagery).

    • API triggers (job submission, credit deduction).

    • Backend processes (Docker containers, GPU allocation).

  • Defined error states (e.g., unsupported bands, credit limits).

2. Console-Centric Design

  • Asset Selection:

    • Table/grid view with batch selection.

    • Clear indicators for process-compatible imagery.

  • Process Discovery:

    • Filterable catalog of algorithms (metadata-driven).

    • Contextual availability (e.g., "Car Detection" only for 50cm resolution).

  • Job Management:

    • Progress tracking + estimated credit cost upfront.

    • Link to SDK for advanced users (AOI/band selection).

3. Validation with Users

  • Tested low-fi prototypes with 5 customers after the 4-week concept phase.

  • Feedback: Prioritized "quick run" defaults over granular controls (for v1).

Impact & Future Scope

  • Short-Term (Q3):

    • Launched core processing flow for full-image analysis.

  • Long-Term:

    • Advanced controls: AOI clipping, band selection.

    • Previews: Visualize outputs before full job execution.

Key Insight:

"Users cared more about ‘what’s possible’ than ‘how it works’—so we surfaced algorithm metadata upfront."