Hibernate

1 / ORM in Practice

Hibernate entered my work through Spring Boot and Spring Data JPA. In projects like Trajectory and Conclave, Hibernate managed the object-relational mapping between Java entity classes and PostgreSQL tables.

2 / The N+1 Reality

The most important lesson Hibernate taught me was about the N+1 query problem. Default lazy-loaded associations generate a separate SQL query for every related entity accessed in a loop. In Trajectory, where application pipeline records had multiple nested associations, this created measurable latency during list queries. The fix was a combination of explicit JPQL fetch joins and, in some cases, abandoning ORM abstractions entirely for native SQL when the query complexity exceeded what JPA could express cleanly.

3 / Where ORM Helps and Where It Hurts

Hibernate excels at mapping straightforward CRUD operations and managing entity lifecycle transitions. It struggles when queries involve complex joins, window functions, or vector distance calculations — which is why projects like Phoenix used raw SQL with pgvector operators instead. The engineering judgment is knowing when ORM abstraction saves time and when it becomes an obstacle to writing the query you actually need.