Qdrant
Rust vector database with payload filtering — strong when metadata matters as much as similarity.
Qdrant's filtering engine is its differentiator: you can constrain by arbitrary payload fields without wrecking latency. That matters more in production than raw ANN benchmarks, because real queries are rarely pure similarity search.
Best for: Filtered vector search and hybrid retrieval
Deploy: Self-hostable
What to do about it
- Check licence terms
- Verify data handling
- Estimate operational cost
What it is
Qdrant is a vector database that allows for filtering by arbitrary payload fields. This means it can handle queries that require both similarity search and metadata constraints.
The filtering engine is a key part of Qdrant, enabling it to constrain search results without significantly impacting latency.
Where it fits in a deployment stack
Vector databases like Qdrant are used to solve the problem of efficient similarity search in high-dimensional data. They typically sit alongside other components such as data ingestion pipelines, indexing systems, and query interfaces.
In a production system, a vector database would normally take over the task of searching for similar items, freeing up other components to focus on tasks like data processing and storage.
By handling similarity search and metadata filtering, a vector database can help to reduce the complexity of queries and improve overall system performance.
How to evaluate it
To evaluate Qdrant, you need to consider how it handles real-world queries and data. This involves testing its performance with your specific use case and data.
- Feed it a dataset with diverse metadata and similarity characteristics.
- Inject failures into the system, such as network partitions or node failures.
- Inspect the output of queries with varying levels of complexity.
- Measure p95 latency of the retrieval step under different concurrency levels.
- Compare the results of similarity searches with and without metadata filtering.
- Verify the correctness of query results using a ground truth dataset.
What to verify before you commit
Before committing to Qdrant, you need to verify several aspects of the tool, including its licence, data handling, and operational cost. You should also check for potential lock-in and failure behaviour.
- Check the licence terms for any restrictions on usage or redistribution.
- Verify how Qdrant handles sensitive data, such as encryption and access controls.
- Estimate the operational cost of running Qdrant, including hardware and maintenance costs.
- Test the failure behaviour of Qdrant, including recovery from node failures or data corruption.
- Check for any potential lock-in, such as proprietary data formats or vendor-specific dependencies.
- Verify the support and community resources available for Qdrant.
Where this fits in the FDE path
- FDE career path — the levels, the titles that map to them, and what changes at each step
- Interview guide — what the loop tests beyond coding
- Pay data — 223 postings — midpoint median $231,000, employer-published ranges only
- 922 live FDE roles — pulled daily from employers’ own job boards
RAG in Production Playbook
Chunking strategies, reranking setups, evaluation harnesses and the seven failure modes that kill every RAG demo in week three.