Wavo.me

The Context

Wavo was a music tech startup building a Music Intelligence Platform (MIP) that helped labels and artists understand their audience. When I joined in October 2018, the data infrastructure was minimal. The platform needed a real cloud architecture and reliable data pipelines before it could deliver on its product vision.

What I Built

I designed and implemented the early cloud architecture for the MIP, building ETL pipelines from scratch in Spark and AWS. The stack included S3 for storage, EMR for processing, RDS for structured data, Glue for cataloging, and Athena for ad-hoc queries. This was a greenfield build: there was no existing data platform when I started.

I also developed time series forecasting systems to prioritize expensive external API calls. The platform relied on third-party data sources that charged per request, so predicting which calls would return the most valuable data saved real money.

What I Learned

This was my first pure data engineering role, and it taught me the value of getting the foundation right before building anything on top of it. The forecasting work was a good example of using ML to solve an operational problem rather than a product problem, a pattern I would see again at Rio Tinto.