Kdrant — an idiomatic, coroutine-first Kotlin client for Qdrant

[tonytonycoder11]

Hi Qdrant team & community 👋

Sharing a project I've been building: Kdrant, an open-source, coroutine-first Kotlin client for Qdrant.

I love using Qdrant from the JVM, but from Kotlin I wanted something that felt native — suspend functions instead of ListenableFuture, a type-safe DSL instead of protobuf builders, and a lighter footprint for typical RAG / embedding-search services. So I built Kdrant on a pure-Kotlin REST engine (Ktor + kotlinx-serialization).

What it gives you:

• Coroutine-first — every operation is a suspend function, with cooperative cancellation & timeouts.
• Type-safe DSLs for collections, points, payloads, and filters.
• Small footprint — no gRPC/Netty/protobuf on the classpath (~3–5 MB added vs ~15–20 MB).
• Typed errors — a sealed KdrantException you can handle exhaustively.

It covers the modern /points/query API — hybrid RRF/DBSF fusion, recommend/discover/context, sparse & multi-vectors, grouped/batch search — plus filters, scroll, aliases, snapshots, and Spring Boot / Spring AI (VectorStore) / LangChain4j (EmbeddingStore) integrations, with a runnable RAG example.

val qdrant = Kdrant(host = "localhost", port = 6333)
qdrant.createCollection("articles") { vector { size = 1_536; distance = Distance.COSINE } }
qdrant.upsert("articles") { point(id = 1) { vector(embedding); payload("title" to "Intro") } }
val hits = qdrant.search("articles") { query(queryVector); limit = 5 }

Honest about trade-offs: for raw throughput and streaming, gRPC still wins — the official Java client is the right call there. Kdrant targets idiomatic Kotlin ergonomics and a small footprint for typical RAG workloads.

• Repo: https://github.com/NaCode-Studios/Kdrant
• Maven Central: io.github.nacode-studios:kdrant-transport-rest:1.1.0
• API docs: https://nacode-studios.github.io/Kdrant/
• License: Apache-2.0

Would love feedback — especially on API ergonomics and anything you'd want from a Kotlin-native client. Thanks for building Qdrant! 🙏