STING Technology Stack - Built on Proven Tools
Built on Technology You Can Trust
No proprietary lock-in. No black boxes. Just proven, enterprise-grade tools.
STING is built on industry-standard, open-source technologies that you already know and trust. Every component is auditable, replaceable, and backed by active communities.
Technology Stack Overview
Modern Frontend
React 18, Material-UI, modern web standards
Scalable Backend
Python/Flask, FastAPI, battle-tested in production
Industry-Standard AI
LangChain, ChromaDB, proven ML tools
Flexible Infrastructure
Docker, Kubernetes, runs anywhere
Auditable Security
HashiCorp Vault, Kratos, standard protocols
Trusted Databases
PostgreSQL, Redis, no vendor lock-in
Detailed Technology Breakdown
Frontend Layer
- React 18 - Modern concurrent rendering, better performance
- Material-UI (MUI) - Professional component library, accessibility built-in
- React Router - Client-side routing
- Redux Toolkit - State management
- React Query - Server state management and caching
- Tailwind CSS - Utility-first styling
- Framer Motion - Smooth animations
- Chart.js / Recharts - Data visualization
- Monaco Editor - Code editing (for API testing)
Backend Layer
- Flask - Lightweight web framework
- FastAPI - High-performance async API framework
- Celery - Distributed task queue for background jobs
- Redis - Caching and message broker
- SQLAlchemy - ORM for database abstraction
- LangChain - LLM orchestration framework
- ChromaDB - Vector database for embeddings
- Hugging Face Transformers - Model loading and inference
- Sentence Transformers - Embedding generation
- Ollama - Local model serving (default)
- vLLM / LM Studio - GPU-accelerated inference (optional)
Data Layer
- PostgreSQL 15+ - Primary relational database
- Redis - In-memory cache and session store
- ChromaDB - Vector database for semantic search
- Minio / S3 - Object storage for files and documents
Pre-built connectors for 15+ data sources:
- Microsoft SQL Server, PostgreSQL, MySQL, Oracle, MongoDB
- Active Directory, Okta, Azure AD
- SharePoint, OneDrive, Google Drive
- Salesforce, HubSpot CRM
- Jira, Confluence, Notion
- Custom REST/GraphQL APIs
Infrastructure Layer
- Docker - Containerization
- Docker Compose - Local/development orchestration
- Kubernetes (optional) - Production-grade orchestration
- Helm Charts - K8s deployment automation
- Nginx - Reverse proxy and load balancing
- Ory Kratos - Identity and authentication
- HashiCorp Vault - Secrets management (Pro+)
- Traefik / Nginx - TLS termination
- Let's Encrypt - Automated SSL certificates
- Fail2ban - Intrusion prevention
AI Model Support
Local Models (Multi-Model Approach)
STING uses different models for different tasks—the right tool for each job. No single model does everything well, so we don’t pretend otherwise.
Qwen 2.5
Best for: General purpose, recommended default
- Multiple sizes (7B, 14B, 32B)
- Excellent all-around performance
- Strong multilingual support
Phi-4 Reasoning
Best for: Complex reasoning & analysis
- 14B parameters
- Strong logical reasoning
- Planning and decision tasks
DeepSeek Coder
Best for: Code generation
- Multiple sizes available
- State-of-the-art coding
- Multi-language support
Multimodal (LLaVA, Qwen-VL)
Best for: Images + text
- Analyze images and documents
- OCR and visual understanding
- Chart and diagram analysis
Cloud AI Integration (Optional)
STING supports hybrid deployment with cloud AI providers:
- OpenAI - GPT-4, GPT-3.5-turbo
- Anthropic - Claude 3 family
- Google - Gemini Pro
- Cohere - Command models
- Any OpenAI-compatible API - Custom endpoints
Policy-based routing lets you control which data uses which models. Sensitive data stays local, complex reports can use cloud AI.
Architecture Diagram
┌─────────────────────────────────────────────────────────┐
│ User Interface │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌─────────┐ │
│ │ Web App │ │ Mobile* │ │ API │ │ CLI* │ │
│ │ (React) │ │ (PWA) │ │ Docs │ │ Tools │ │
│ └──────────┘ └──────────┘ └──────────┘ └─────────┘ │
└─────────────────────────────────────────────────────────┘
│
┌─────▼─────┐
│ Nginx │
│ (Proxy) │
└─────┬─────┘
│
┌─────────────────────────▼───────────────────────────────┐
│ Application Layer │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌─────────┐ │
│ │ Flask │ │ FastAPI │ │ Celery │ │ Bee AI │ │
│ │ (API) │ │ (Async) │ │ (Tasks) │ │ (Agent) │ │
│ └──────────┘ └──────────┘ └──────────┘ └─────────┘ │
└─────────────────────────────────────────────────────────┘
│
┌─────────────────────────▼───────────────────────────────┐
│ Data Layer │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌─────────┐ │
│ │Postgres │ │ Redis │ │ ChromaDB │ │ Minio │ │
│ │ (SQL) │ │ (Cache) │ │ (Vector) │ │ (S3) │ │
│ └──────────┘ └──────────┘ └──────────┘ └─────────┘ │
└─────────────────────────────────────────────────────────┘
│
┌─────────────────────────▼───────────────────────────────┐
│ AI Layer │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌─────────┐ │
│ │ Ollama │ │ Cloud AI │ │ Custom │ │ vLLM │ │
│ │ (Local) │ │(Optional)│ │ Models │ │ (GPU) │ │
│ └──────────┘ └──────────┘ └──────────┘ └─────────┘ │
└─────────────────────────────────────────────────────────┘
*Coming in future releases
Development Tools
- Python 3.11+ - Modern Python features
- Node.js 18+ - JavaScript runtime
- TypeScript - Type-safe frontend development
- ESLint / Prettier - Code quality and formatting
- Black / Flake8 - Python code formatting and linting
- Pytest - Python testing framework
- Jest / React Testing Library - Frontend testing
- GitHub Actions - CI/CD pipelines
Why These Technologies?
✅ Battle-Tested in Production
Every technology in the STING stack has been proven at scale by thousands of companies. No experimental tech, no untested frameworks.
🔓 Open Source & Auditable
All core technologies are open source. Your security team can audit every line of code. No proprietary lock-in.
👥 Active Communities
Each technology has a large, active community. Finding developers is easy. Getting help is easy. Future support is guaranteed.
Easy to Replace
Standard interfaces mean you can swap components. Don't like PostgreSQL? Use MySQL. Prefer Kubernetes? We support that too.