C

backend technical write-up

PromptParty

Graph-backed prompt collaboration platform

PromptParty explores how prompt versions, users, sessions, feedback, and reusable context can be modeled as connected data instead of flat records.

All projectsGitHubLive site

Problem / motivation

Prompt-heavy tools quickly become hard to reason about: prompts branch, teams reuse snippets, evaluations change, and performance can suffer when every relationship is reconstructed at request time.

Key technical challenges

  • Balancing graph flexibility with predictable API contracts.
  • Keeping relational and graph data responsibilities clear.
  • Designing queries that stay fast as prompt histories grow.

Architecture / workflow

How the system fits together

01

FastAPI service receives authenticated requests and normalizes prompt workflow operations.

02

GraphQL layer exposes flexible read shapes for project, prompt, and relationship views.

03

Neo4j stores prompt lineage, relationships, and graph traversal paths.

04

Supabase/PostgreSQL handles account data, stable relational records, and metadata.

05

Performance passes focus on query shape, indexing, and avoiding repeated graph traversals.

What I built

  • Data model for prompt versions, relationships, and collaboration artifacts.
  • API structure around prompt lifecycle operations and graph-backed exploration.
  • Performance-minded query patterns for common traversal paths.

Outcomes / metrics

  • Clearer architecture for a product that mixes collaborative workflows and graph data.
  • A strong backend project write-up showing API design, database boundaries, and performance thinking.

Lessons learned

  • Graph databases are powerful when relationships are first-class, but the boundary with relational storage has to be explicit.
  • API design should protect product velocity from underlying data-store complexity.

Screenshots / media

Visual evidence placeholders

Replace these panels with screenshots, demos, diagrams, or notebook exports as each artifact becomes ready for publishing.

Media

Graph workflow model

Prompt nodes, version edges, feedback loops, and workspace metadata.

Media

API boundary

FastAPI coordinates GraphQL reads, relational writes, and graph traversal logic.