asterinis

Asterinis

Asterinis is an early-stage Python framework for connecting NLP, retrieval-augmented generation (RAG), AI agents, routing logic, explainability, and LLM services through a unified backend nexus.

The project is designed as a lightweight orchestration layer that can connect different AI and NLP components without locking an application to a single model provider or framework.

Current Status

Asterinis is currently in an early development stage.

The first public release focuses on the core framework structure, including:

• Nexus orchestration
• Rule-based routing
• Connector interfaces
• Processing context
• Pipelines
• Hooks
• Result objects
• Optional NLP integrations

More advanced RAG, agent orchestration, adaptive routing, evaluation, and explainability features are planned for future releases.

Installation

The package will be installable from PyPI:

pip install asterinis

For optional Flair NLP support:

pip install "asterinis[flair]"

Basic Usage

from asterinis import Nexus

nexus = Nexus()

result = nexus.process(
    "Search documents about multilingual NLP."
)

print(result.route)
print(result.payload)

Asterinis can route requests toward different processing paths such as:

rag
nlp
agent
llm

Flair Integration

Asterinis supports Flair as an optional NLP connector.

from asterinis import Nexus
from asterinis.integrations import FlairConnector

nexus = Nexus()

nexus.register_connector(
    "nlp",
    FlairConnector("ner")
)

result = nexus.process(
    "Deutsche Bank is based in Frankfurt."
)

print(result.to_dict())

Flair is an independent open-source NLP framework and is not part of Asterinis. Asterinis provides an integration layer that allows Flair models to participate in a broader orchestration pipeline.

Architecture

Application
    ↓
Asterinis Nexus
    ↓
Router
    ↓
┌──────────┬──────────┬──────────┬──────────┐
│   NLP    │   RAG    │  Agents  │   LLM    │
└──────────┴──────────┴──────────┴──────────┘
    ↓
Connectors / Integrations

The goal is to allow developers to plug different AI systems into a consistent interface.

Project Structure

src/
└── asterinis/
    ├── __init__.py
    ├── nexus.py
    ├── router.py
    ├── connectors.py
    ├── context.py
    ├── pipeline.py
    ├── hooks.py
    ├── result.py
    ├── config.py
    ├── exceptions.py
    └── integrations/
        ├── __init__.py
        └── flair.py

Roadmap

Future development may include:

• advanced RAG routing
• retrieval quality evaluation
• agent orchestration
• adaptive routing policies
• explainability
• confidence-aware decision logic
• additional NLP integrations
• LLM provider connectors
• evaluation tools
• persistent context and memory
• asynchronous pipelines

Version

Current development version:

0.0.1

License

Asterinis is released under the MIT License. See the LICENSE file for details.

Author

Created and maintained by Irena Popova, Full-Stack Developer and AI/NLP, RAG, and adaptive systems researcher.