Next 
Val Data
Data insights, analytics, and technical explorations.
I’m Dominick Ryan. As a Senior Systems and Data Engineer, I build reliable data transformations and cloud-native infrastructure. NextValData is my curated repository of academic and technical documentation—a personal archive shared in the hope that it helps others, while being just self-serving enough.
Latest Posts
In my previous post, I detailed the 'how' of unpivoting 21 million rows of genomic data using DuckDB. Since then, the project has matured significantly. I’ve implemented Dagster for orchestration and moved the compute to a Spark-on-Delta Lakehouse architecture.
Understanding how Large Language Models (LLMs) function is heady stuff. The difficulty lies partially in the sheer velocity of the field, and partially because our metaphors haven’t yet caught up to the methodology. But this isn't just a technical shift; it's a new climate of opportunity and challenge for how we manage the flow of intelligence.
Transforming a 21-million-row matrix into a normalized relational structure for a FAIR data environment.