[Enhancement]: Develop backend for AI-driven summarization of simulation metadata

[tomvothecoder]

Is your feature request related to a problem?

Researchers currently have to manually interpret and summarize structured simulation metadata, which is time-consuming and prone to inconsistency. There’s no automated way to generate concise summaries that highlight key parameters, scientific goals, and relationships between simulations within SimScope.

Describe the solution you'd like

Implement a FastAPI-based backend module that uses transformer-based models to automatically generate structured, human-readable summaries from simulation metadata.

Key capabilities:

• Parse simulation metadata (JSON) submitted through the SimScope frontend.

• Use AI/LLM-based summarization pipelines (e.g., OpenAI or local transformers) to:

  • Produce concise textual summaries.
  • Identify similarities/differences between simulations.
  • Highlight key metadata fields or anomalies.

• Expose REST endpoints such as:

  • POST /analyze → returns AI-generated summary.
  • POST /compare → returns comparative summary between simulations.

• Provide a modular design for switching AI models and prompts.

Describe alternatives you've considered

N/A

Additional context

This feature supports SimScope’s goal of surfacing key insights from E3SM and related Earth system simulations.

Integration targets:

• Existing FastAPI backend and Pydantic metadata models.
• Future database persistence for caching AI summaries.
• Alignment with E3SM Simulation Viewer and metadata submission workflows.