AIONBD
Deploy vector search on branch sites, IoT devices ...
Start new thread
Ayoubn Nabil

AIONBD - Deploy vector search on branch sites, IoT devices ...

by•
AIONBD, Vector Database for the Edge Deploy AI vector search where cloud won't work. 12x faster than Qdrant. Single binary. Production-ready. — BENCHMARKS (Fashion-MNIST 784-dim, top-k=10) AIONBD: 767 QPS, p95: 1.4ms ⚡ Qdrant: 65 QPS, p95: 21ms Same hardware. Same test.

Add a comment

Replies

Best
Ayoubn Nabil
Maker
📌
I built AIONBD because I needed a vector database strictly for edge environments. The current major engines (like Qdrant or Chroma) are great, but they are bloated and will gladly eat up 2GB of RAM just sitting idle. I engineered AIONBD to have a strictly enforceable memory budget. You can set AIONBD_MEMORY_BUDGET_MB=4, and it will reliably run on 4MB of RAM without crashing, rejecting inputs with a clear resource_exhausted error when full. Or you can give it 4000MB and it will scale linearly.