ObjectBox Java – On‑device NoSQL database with vector search

What it is

ObjectBox‑Java is a lightweight, on‑device database for Java and Kotlin apps (Android, desktop, server). It stores plain objects (POJOs) directly, so you never write SQL. The library also includes a vector‑search engine, letting you keep high‑dimensional embeddings alongside your regular data and run fast similarity queries – useful for RAG, generative‑AI or recommendation features.

Why you might want it

Feature	What it means for you
First on‑device vector DB	Store embeddings locally and run similarity search without a remote service.
High performance	Benchmarks claim CRUD operations are faster than SQLite or Realm, with low CPU, memory and battery impact – important for mobile or edge devices.
Object‑oriented API	Define an @Entity class, get a Box<T> and call put, get, remove. No SQL strings, compile‑time type safety.
Built‑in relations	One‑to‑many, many‑to‑many links are handled automatically, simplifying complex data models.
Cross‑platform	Works on Android and any JVM (Linux, macOS, Windows) and supports ARM CPUs.
Sync & time‑series extensions (separate products)	Optional add‑ons for syncing data between devices or storing time‑series data.

How you use it (quick sketch)

@Entity
public class Person {
    @Id long id;               // auto‑generated primary key
    String firstName;
    String lastName;
}

BoxStore store = MyObjectBox.builder()
        .name("person-db")
        .build();
Box<Person> box = store.boxFor(Person.class);

Person p = new Person();
long id = box.put(p);          // create
Person loaded = box.get(id);   // read
loaded.lastName = "Smith";
box.put(loaded);               // update
box.remove(loaded);            // delete

Kotlin usage is analogous, using the same annotations.

Getting it into your project

• Gradle – add the io.objectbox plugin (or use a version catalog) and the ObjectBox dependencies from Maven Central.
• Maven – supported for JVM projects; see the example repo for a pom file.
• Requires Java 8+ (or Kotlin 1.7+), runs on Android 5.0+ and on modern desktop OSes.

Where to learn more

• Docs & getting‑started – https://docs.objectbox.io
• Example apps – https://github.com/objectbox/objectbox-examples
• FAQ / troubleshooting – linked from the README.
• Performance benchmark repo – shows speed vs. SQLite/Realm.

License

Apache 2.0 (the code in this repo). Other ObjectBox components have their own licensing terms.

---

ObjectBox Java is a genuine software project focused on local data storage and on‑device vector search, fitting the AI/ML frontier by enabling edge‑side similarity search for embeddings.