Coursework Detail Page
Charts Isle
An early data-focused full-stack project exploring song popularity by country with CSV-to-MongoDB ingestion pipelines.
Hero
A full-stack music exploration app that maps popular Spotify songs to the countries where they trended, built as an early team project and later refined for portfolio quality.
The application lets users search songs and explore where those songs were popular across different countries. It started as a school collaboration focused on connecting data engineering and web development fundamentals, then became a strong portfolio piece because it required practical database design, ingestion workflows, and API-to-UI integration.
My work centered on backend and data concerns: shaping MongoDB collections to support country lookups and song exploration, integrating those models through Express API endpoints, and building the ingestion pipeline that transformed raw CSV inputs into cleaner, queryable documents. I also worked on search behavior and response design so frontend interactions stayed predictable while querying large collections.
Source datasets were ingested with Python and pandas, where parsing and cleanup happened before load into MongoDB. The final dataset included roughly 521k country records, 245 location records, and around 10k music records. Collection relationships connected song entries with country and location metadata, making it possible to support country-based discovery and cross-reference flows in a single user journey. The full stack path was intentionally end to end: transformed data in MongoDB, API access through Node/Express with Mongoose models, and React components rendering searchable results. Users could also submit feedback comments that were written back to music documents.
A key complexity was balancing geospatial-style country exploration and text-based lookup performance. Query behavior had to be tuned to avoid slow runtime scans, especially when combining song search intent with country metadata.
The hardest part was data quality. Raw CSV inputs contained noisy and inconsistent values, so ingestion needed stronger normalization than expected before records were safe to query at scale. I also learned how quickly search performance can degrade in semi-structured NoSQL datasets when query patterns are not designed early. This project made indexing strategy, schema discipline, and response-shaping a first-class concern rather than an afterthought.
If I rebuilt this today, I would add stricter ingestion contracts, clearer incremental ETL boundaries, and more explicit query observability to measure runtime hotspots before they affect the user experience.
This project was an early turning point in my development. It reinforced that I enjoy backend API design and data platform work more than purely presentation-layer development, and it gave me practical experience with NoSQL modeling, data transformation, and full-stack integration under realistic constraints. That foundation directly influenced my continued focus on backend and data engineering projects.