A lightweight ODM for Qdrant — looking for feedback

I’ve been working with Qdrant in a few projects, and as things scaled, I found myself repeatedly dealing with:

• raw payload field strings
• verbose filter construction
• scattered schema and index management

To make this easier to work with, I built a small ODM-style layer on top of Qdrant.

---

What this tries to solve

The goal is not to hide Qdrant, but to make it more structured and easier to maintain:

• define schema and indexes at the model level
• write filters using Python expressions
• keep CRUD and search logic consistent

---

Example

class Document(QdrantModel):
    __collection__ = "documents"

    title: str = PayloadField(index="keyword")
    page: int = PayloadField(index="integer")

(Document.category == "tech") & (Document.page >= 1)

---

Repo

https://github.com/Jeung-SeongYeon/qdrant-odm

---

Feedback

This is still an early version (v0.3.x), and I’m actively refining the design.

If you have experience with Qdrant or similar systems, I would really appreciate any feedback on:

• API design
• missing features
• overall usability

Also, if you find it useful, a ⭐ would help a lot.

Thanks in advance.