Aithena

Aithena is the umbrella repository for all Aithena projects, providing a complete ecosystem for building advanced AI services, specifically focused on scientific research assistance and RAG (Retrieval-Augmented Generation) systems.

This monorepo contains the agents, applications, services, and deployment configurations required to run the full stack.

🏗️ Architecture & Documentation

We have detailed architectural documentation available in the diagrams/ directory.

• AskAithena Structure: A detailed guide to the AskAithena platform architecture, detailing services, their roles, and interactions.
• System Architecture: A high-level overview of the entire Aithena ecosystem on the Polus servers.

Note: For visual diagrams, see the .svg files in the diagrams/ directory (e.g., AskAithena_Structure.svg).

📂 Project Structure

Directory	Description
agents/	AI Agents and logic. Contains the Ask Aithena Agent, the core RAG intelligence.
apps/	User-facing applications. Contains the Ask Aithena Web App, a modern Next.js interface.
deployments/	Infrastructure as Code. Helm charts, Kubernetes manifests, and Slurm scripts.
embeddings/	Shared libraries for embedding generation and management.
jobs/	Data ingestion and processing jobs (e.g., fetching OpenAlex data).
mcp/	Model Context Protocol implementations (e.g., Ask Aithena MCP).
services/	Backend microservices. Includes Aithena Services for vector memory and LLM integration.
templates/	Project templates for creating new services or agents.

🚀 Core Components

1. Ask Aithena Agent

The brain of the operation. A sophisticated RAG system that answers scientific questions based on 150+ million academic articles.

• Key Features: Multi-agent architecture (Semantic Extractor, Retriever, Responder), 3 protection levels (Owl, Shield, Aegis), and Pydantic AI integration.

2. Ask Aithena App

The face of the platform. A Next.js web application that provides a chat interface for users to interact with the agent.

• Key Features: Real-time streaming, RabbitMQ updates, and a responsive UI.

3. Aithena Services

The backbone. A service layer handling vector memory (pgvector), database functionality, and LiteLLM integration.

• Key Features: Unified API for LLMs and memory, optimized vector search.

🛠️ Deployment

All deployment configurations are located in the deployments/ directory.

Kubernetes (deployments/kubernetes/)

The deployments/kubernetes/ directory contains manifests for deploying individual services.

IMPORTANT: Each subdirectory in deployments/kubernetes/ (e.g., postgres, litellm, ask-aithena-app) contains its own README.md with specific deployment instructions, configuration details, and prerequisites. Please consult these individual READMEs before deploying.

• Postgres: Database configuration and initialization.
• LiteLLM: LLM Gateway deployment.
• Ask Aithena App: Frontend deployment.
• ...and others.

Helm (deployments/helm/)

Helm charts for packaging and deploying the applications.

Slurm (deployments/slurm/)

Scripts for running jobs in HPC environments (Singularity/Apptainer).