SoundHash

SoundHash is a Python application for identifying short audio clips by fingerprinting video audio and matching it against previously ingested YouTube content. I built it to demonstrate the kind of backend and data-engineering work I enjoy most: turning messy media inputs into a reliable, searchable system.

Why it stands out

• Designed an end-to-end ingestion and processing pipeline for YouTube audio
• Applied digital signal processing techniques to generate compact audio fingerprints
• Built backend services with Python, FastAPI, SQLAlchemy, and PostgreSQL
• Structured the project for real-world operation with Docker, environment-based config, logging, and background job tracking
• Added automated testing, linting, type checking, and CI workflows to support maintainability

Technical highlights

• Backend engineering: Python 3, FastAPI, SQLAlchemy, Alembic
• Data & storage: PostgreSQL, repository pattern, processing job lifecycle management
• Media processing: yt-dlp, ffmpeg, spectral analysis, fingerprint hashing
• Platform/API work: YouTube integration, OAuth flows, REST endpoints, auth and rate limiting
• Developer experience: pytest, Ruff, Black, mypy, Docker, GitHub Actions

What this project demonstrates

This project showcases my ability to:

• design and ship multi-stage processing pipelines
• work across APIs, databases, and media-processing tools
• translate technical complexity into maintainable application architecture
• build software with production-minded practices rather than one-off scripts

Quick look

python scripts/setup_database.py
python scripts/ingest_channels.py --dry-run --max-videos 5

Repository structure

• src/core/ – audio fingerprinting and video processing
• src/ingestion/ – orchestration for channel ingestion and processing jobs
• src/database/ – database models, connections, and repositories
• src/api/ – YouTube integration and API-related services
• config/ – centralized settings and logging configuration

Notes for reviewers

If you're evaluating this project as part of a job application, the main strengths to look for are system design, backend implementation, data pipeline thinking, and practical use of Python tooling around a non-trivial media problem.