About the Project

From a Big Data initiative to a full-scale deep learning research study.

This project began as a Big Data and Data Engineering initiative and evolved into a full-scale research study on music recommendation systems. Over a span of four months, it investigated the end-to-end lifecycle of music platforms—focusing on how raw choices in data handling impact recommendation quality.

The goal is to bridge data engineering practices with machine learning research, demonstrating how low-level decisions—like ETL pipeline design and acoustic feature extraction—directly influence the performance of high-level recommendation models.

Core Research Areas

  • Massive audio dataset management
  • Distributed ETL pipelines for metadata
  • Feature extraction from raw waveforms
  • Deep learning models for music understanding

Functional Integrations

  • Raw Audio (WAV) Processing
  • Spectral Feature Analysis
  • Lyrics & Sentiment Pipelines
  • Deep Audio Embeddings
  • Music Metadata (MusicBrainz)