socra AI Tech Stack
TL;DR: Socra is built with a Django backend and Next.js frontend, utilizing a relational database vi
TL;DR: Socra is built with a Django backend and Next.js frontend, utilizing a relational database via Postgres, containerization with Docker, CI/CD with GitHub Actions, and container management and deployment with Kubernetes and Helm. All of this supports a comprehensive user experience. :)
### Backend: Django via API
- **LLMs for generative text**: OpenAI API, Anthropic, etc
- **Embeddings for personalization via RAG**: OpenAI Ada2
- **Image generation for interactivity with users**: Stable Diffusion
- **Payments**: Stripe
- **Core features/libraries**:
- Auth & user management
- Blocks (rich text editor)
- Feature flags & feature gating
- Socras, Chats, Journeys
- Fully in-house custom email system with ad campaigns & analytics
- Platform analytics system
- Websocket for async communications with clients
- User settings
- **Asynchronous task queue**: Django Celery, Celery Beat with Redis integration
- **Error tracking**: Sentry
### Database: Postgres
- Utilizes the pgVector add-in for in-house embeddings, allowing for faster retrieval and segmentation.
### Client: Next.js (React)
- **Main libraries used**: MUI, SWR, Lexical, and client-side libraries for the aforementioned features
- **Monorepo structure**
### CI/CD: GitHub Actions
- Complete pipeline for formatting, linting, testing, and building the containers our app uses
- Automatic deployment to our clusters
### Cloud, Servers, etc.
- **AWS** for cloud hosting (Route 53, ECS, EKS, RDS, etc.)
- Containerization of backend, frontend, documentation, etc.
- Kubernetes clusters for development and production environments (automatic deployment via CI/CD), managed with Helm
Let us know what you think!By Eduarda Ferreira