Youssef

nexary-platform

On-premise enterprise AI platform: multi-provider LLM chat, hybrid RAG (dense + BM25 + cross-encoder rerank), SYSLOG-auditable, runs inside customer networks

Code on GitHub

Language
TypeScript
Licence
MIT
Last push
30 Oct 2026
Stars
1

What it is

An on-premise AI platform export for multi-provider chat, document retrieval and auditable operations. Provider egress is optional configuration; the deployment keeps application data in customer-controlled infrastructure.

How it works

Regulated teams need to inspect where data goes, retrieve exact identifiers as well as concepts, and export activity to existing operations tooling.

Status: public portfolio export. The production Linux deployment was previously verified; this export has not been re-verified as a fresh end-to-end private-cloud install. The OCR service is separate, and no customer data or credentials are included.

Results

Control Retrieve Operate
Customer-network deployment and SSO Dense retrieval + BM25 + reranking Audit events and SYSLOG export
Docker Compose, PostgreSQL and MinIO Qdrant and PostgreSQL full text Health endpoints and runbooks

Try it

cp .env.example .env
docker compose -f docker-compose.private-cloud.yml up -d --build