Shelf
Unified tracker for media collections with external metadata sync.
Overview
Shelf helps users organize shows, movies, anime, and books with consistent metadata and progress tracking.
The backend aggregates multiple public APIs and normalizes fields for a clean front-end experience.
Problem
Media tracking tools often focus on one domain and become fragmented across apps.
Metadata quality and naming conventions vary between providers, creating messy search and duplicate entries.
- —Cross-source fields use different formats and confidence levels.
- —Users need a single progress model across multiple media types.
- —Deployment must remain reproducible and easy to update.
Constraints
API heterogeneity
TMDB, Jikan, and Open Library return different schemas and quality guarantees.
Data freshness
The system needed periodic synchronization without degrading app response times.
Operational simplicity
Containerized deployment and CI/CD had to stay simple enough for quick iteration.
Key Decisions
Normalized metadata layer
Created a mapping layer that translates provider-specific payloads into a single internal model.
Service split by responsibility
Separated ingestion, persistence, and user collection features to keep core workflows testable.
Container-first deployment
Used Docker and pipeline automation for consistent environment setup and safer releases.
Results
- —Delivered a unified tracking experience across different media categories.
- —Reduced duplicate metadata noise through normalization rules.
- —Enabled reliable deployments with automated build and release flow.
Takeaways
- —Data normalization is the core product feature when aggregating multiple providers.
- —Clear service boundaries make API-heavy apps easier to evolve.
- —Operational tooling early in the project increases long-term velocity.