Unified platform for self-hosted LLM inference + enterprise safety governance
Project description
TurboPrivate AI — Self-Hosted Enterprise AI Platform
Switch from OpenAI in 30 seconds. Drop-in compatible API with built-in safety, governance, and 40–60% cost reduction.
Run powerful LLMs on your own hardware — with enterprise safety, governance, and full data sovereignty.
Quick Start
One-Click Install
curl -fsSL https://get.turboprivate.ai | bash
Or via pip
pip install turboprivate-ai
turbo deploy --provider bare-metal --gpu auto
turbo model serve meta-llama/Llama-3.1-8B --quant int4
turbo chat
Docker Compose
git clone https://github.com/Kubenew/turboprivate-ai.git
cd turboprivate-ai
docker compose -f docker-compose.full.yml up -d
Why TurboPrivate AI?
| Feature | TurboPrivate AI | Ollama | vLLM | OpenAI API |
|---|---|---|---|---|
| Data Sovereignty | ✅ Full | ✅ Full | ✅ Full | ❌ Cloud |
| Enterprise Safety | ✅ Mythos Safe (7 verifiers) | ❌ None | ❌ None | ⚠️ Basic |
| OpenAI Compatible | ✅ 100% | ✅ Partial | ✅ Partial | ✅ Native |
| INT4/AWQ Quantization | ✅ TurboQuant v3 | ✅ GGUF | ✅ AWQ | N/A |
| RAG Pipeline | ✅ Built-in | ❌ External | External | ❌ External |
| Audit Trail | ✅ Immutable JSONL | ❌ None | ❌ None | ⚠️ Limited |
| RBAC / Multi-tenant | ✅ Enterprise | ❌ None | ❌ None | ✅ Enterprise |
| Kubernetes Native | ✅ Helm + K3s | ❌ Manual | ⚠️ Manual | N/A |
| Cost (RTX 4090) | ~8x cheaper | Free | Free | $5-10/M tokens |
🏢 For Enterprises
TurboPrivate AI is built for organizations that need control, compliance, and cost efficiency:
Security & Compliance
- Full data sovereignty: Nothing leaves your infrastructure
- Mythos Safe: 7-layer defense (injection, PII, toxicity, hallucination, etc.)
- Audit trail: Immutable JSONL logs with SIEM integration
- RBAC: Fine-grained access control with OIDC/SAML support
- Compliance ready: GDPR, HIPAA, SOC 2, PCI-DSS, ISO 27001
See SECURITY.md and docs/COMPLIANCE.md for details.
Enterprise Integrations
- SAP HANA: Vector store + RAG pipeline (Guide)
- SAP AI Core: BYOM deployment support
- Kubernetes: Helm charts, HPA, multi-cluster
- Observability: Prometheus, Grafana, OpenTelemetry
- Secrets: HashiCorp Vault, AWS Secrets Manager, K8s Secrets
Support & SLAs
| Tier | Response | Includes |
|---|---|---|
| Community | GitHub Issues | OSS core, docs, community support |
| PoC / Pilot | 48h | 4-8 week trial, 2 models, training |
| Enterprise | 4h | SLA 99.5%, unlimited models, TAM |
| Enterprise Plus | 1h | Multi-cluster, custom verifiers, SOC2 |
📅 Book a 30-min PoC Call | ✉️ Contact Sales
📊 Performance (RTX 4090)
| Model | Quant | Tokens/sec | VRAM | Cost vs Cloud |
|---|---|---|---|---|
| Llama 3.1 8B | INT4 | 110+ | ~5.8 GB | ~8x cheaper |
| Qwen2.5 32B | INT4 | 45+ | ~22 GB | ~6x cheaper |
| Llama 3.1 70B | INT4 | 18+ | ~48 GB | ~5x cheaper |
Independent benchmarks: benchmarks/
🛡️ Architecture
CLI / SDK / Dashboard
↓
API Gateway (FastAPI · Auth · Rate Limiting)
↓
┌─────────────────┐ ┌───────────────────┐
│ Mythos Safe │ │ TurboQuant INT4 │
│ Verifiers · │ │ vLLM/llama.cpp │
│ Audit Trail │ │ Inference Engine │
└─────────────────┘ └───────────────────┘
↓
Memory & RAG (TurboMemory · pdf2struct)
↓
──────────┐ ┌──────────┐ ┌──────────┐
│ K3s │ │Monitoring│ │ Storage │
│ Cluster │ │Prom/Graf │ │ PG/Redis │
└────────── └──────────┘ └──────────┘
🎬 Demo
Documentation
- Architecture — System design
- Deployment — Production guide
- Enterprise Guide — Air-gapped, HA, sizing, migration
- Compliance — GDPR, HIPAA, SOC 2, PCI-DSS readiness
- SAP HANA Integration — Cost calculator, security checklist
- CLI Reference — All commands
- API Reference — FastAPI routes
- Security Policy — Vulnerability reporting
- Contributing — How to contribute
🔄 Changelog
0.1.7 (2026-05-17)
- SECURITY.md with threat model, hardening guide, SBOM, responsible disclosure
- CONTRIBUTING.md with dev setup, testing, PR guidelines
- Enterprise Deployment Guide: air-gapped, HA, secrets, proxy, hardware sizing
- Compliance readiness: GDPR, HIPAA, SOC 2, PCI-DSS, ISO 27001, EU AI Act
- One-click installer (install.sh) + docker-compose.full.yml with GPU passthrough
- GitHub issue templates: bug report, feature request, security report
- README overhaul: feature comparison table, "For Enterprises" section, badges
0.1.6 (2026-05-16)
- SAP HANA integration guide: cost calculator, security checklist, BYOM, compliance
- Enterprise hardening best practices
- SECURITY.md and CONTRIBUTING.md added
0.1.5 (2026-05-16)
- SAP HANA vector store integration (LangChain + HanaDB)
- FastAPI RAG endpoint with similarity search
- Document ingestion with PDF/text + HNSW index
0.1.4 (2026-05-13)
- Production Helm charts (configmap, ingress, services)
- TurboQuant v3: AWQ + INT4 mixed-precision
- K3s provisioner with multi-node discovery
- vLLM backend: speculative decoding + prefix caching
📄 License
Apache 2.0 — see LICENSE.
Built by Kubenew — ex-HPE engineer, 12+ years enterprise infrastructure
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file turboprivate_ai-0.1.7.tar.gz.
File metadata
- Download URL: turboprivate_ai-0.1.7.tar.gz
- Upload date:
- Size: 1.1 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.10
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
9a88a987220be0cc465912ce5f9218a328b3ec16aa25870763f13d835d64e62a
|
|
| MD5 |
532eecf17a2e9368d65336e0f8ccdbff
|
|
| BLAKE2b-256 |
19cc8ec3acbd5a01cbdbb9867e9f5af1793274b21b9208a413c278cfbb79e961
|
File details
Details for the file turboprivate_ai-0.1.7-py3-none-any.whl.
File metadata
- Download URL: turboprivate_ai-0.1.7-py3-none-any.whl
- Upload date:
- Size: 55.2 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.10
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
34c152b169c01d447be9fd5217c545f5c9eac7e38cb8e0d8fd70531e8436e0e4
|
|
| MD5 |
11b58550d9366a89a23e820773f5f13f
|
|
| BLAKE2b-256 |
aec44f19f70be2a70dc7909f7aea19c2880621a11a48b4ea96cb5a79166d0c77
|