Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.18983870 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901066564501504 |
|---|---|
| author | Rithan Nandakumar |
| author_facet | Rithan Nandakumar |
| contents | <h2>QueryRouter-AI v1.0.0</h2> <p>Initial public release of <strong>QueryRouter-AI</strong>, a modular AI query routing system that dynamically selects the most appropriate execution path (LLM, search, or hybrid) using a graph-based orchestration workflow.</p> <p>The project demonstrates structured <strong>LLM orchestration using LangGraph</strong>, avoiding monolithic prompt chains and enabling transparent, debuggable execution pipelines.</p> <h2>Key Features</h2> <ul> <li><strong>Graph-based query routing</strong> using LangGraph</li> <li><strong>Intent-aware routing node</strong> that determines the appropriate execution path</li> <li><strong>Groq-powered LLM inference</strong> for fast model responses</li> <li><strong>Optional search pipeline</strong> using SerpAPI for retrieval-augmented responses</li> <li><strong>Streamlit-based interactive UI</strong> for testing and demonstration</li> <li><strong>Modular node architecture</strong> for extensibility and experimentation</li> </ul> <h2>System Architecture</h2> <p>QueryRouter-AI follows a structured execution pipeline:</p> <ol> <li><p><strong>User Query</strong> is submitted through the Streamlit interface</p> </li> <li><p><strong>Router Node</strong> analyzes the query and selects the execution path</p> </li> <li><p><strong>Execution Path</strong></p> <ul> <li>Direct <strong>LLM response</strong></li> <li><strong>Search → Fetch → LLM reasoning</strong> pipeline</li> </ul> </li> <li><p><strong>Final Answer</strong> returned to the UI</p> </li> </ol> <p>This graph-based design improves observability, debugging, and extensibility compared to traditional single-prompt LLM applications.</p> <h2>Project Structure</h2> <pre><code>QueryRouter-AI/ │ ├── ui.py # Streamlit interface ├── Groq.py # LangGraph graph construction and execution ├── Search_fetch.py # Search and retrieval pipeline (SerpAPI) ├── flow.png # Architecture diagram ├── docker-compose.yml # Optional container setup ├── requirement.txt # Python dependencies └── README.md </code></pre> <h2>Technology Stack</h2> <ul> <li>Python 3.10</li> <li>LangGraph</li> <li>LangChain Core 0.2.x</li> <li>Groq LLM API</li> <li>SerpAPI</li> <li>Streamlit</li> <li>FAISS / RAG-ready architecture</li> </ul> <h2>Setup</h2> <pre><code>git clone https://github.com/Rithan377/QueryRouter-AI.git cd QueryRouter-AI python3.10 -m venv venv source venv/bin/activate pip install -r requirement.txt </code></pre> <p>Required API keys:</p> <ul> <li>Groq API Key</li> <li>SerpAPI Key</li> </ul> <h2>Purpose</h2> <p>QueryRouter-AI demonstrates <strong>structured LLM orchestration</strong> where execution paths are explicitly defined through a graph rather than relying on single-call prompt logic. This approach enables easier debugging, modular experimentation, and more production-aligned AI workflows.</p> <h2>Future Improvements</h2> <ul> <li>Advanced routing strategies</li> <li>Memory-aware query handling</li> <li>Vector database integration</li> <li>API-only (headless) deployment mode</li> <li>Evaluation and observability tooling</li> </ul> <h2>License</h2> <p>MIT License</p> <p><strong>Full Changelog</strong>: https://github.com/Rithan377/QueryRouter-AI/commits/v1.0.0</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18983870 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Rithan377/QueryRouter-AI: QueryRouter-AI Rithan Nandakumar <h2>QueryRouter-AI v1.0.0</h2> <p>Initial public release of <strong>QueryRouter-AI</strong>, a modular AI query routing system that dynamically selects the most appropriate execution path (LLM, search, or hybrid) using a graph-based orchestration workflow.</p> <p>The project demonstrates structured <strong>LLM orchestration using LangGraph</strong>, avoiding monolithic prompt chains and enabling transparent, debuggable execution pipelines.</p> <h2>Key Features</h2> <ul> <li><strong>Graph-based query routing</strong> using LangGraph</li> <li><strong>Intent-aware routing node</strong> that determines the appropriate execution path</li> <li><strong>Groq-powered LLM inference</strong> for fast model responses</li> <li><strong>Optional search pipeline</strong> using SerpAPI for retrieval-augmented responses</li> <li><strong>Streamlit-based interactive UI</strong> for testing and demonstration</li> <li><strong>Modular node architecture</strong> for extensibility and experimentation</li> </ul> <h2>System Architecture</h2> <p>QueryRouter-AI follows a structured execution pipeline:</p> <ol> <li><p><strong>User Query</strong> is submitted through the Streamlit interface</p> </li> <li><p><strong>Router Node</strong> analyzes the query and selects the execution path</p> </li> <li><p><strong>Execution Path</strong></p> <ul> <li>Direct <strong>LLM response</strong></li> <li><strong>Search → Fetch → LLM reasoning</strong> pipeline</li> </ul> </li> <li><p><strong>Final Answer</strong> returned to the UI</p> </li> </ol> <p>This graph-based design improves observability, debugging, and extensibility compared to traditional single-prompt LLM applications.</p> <h2>Project Structure</h2> <pre><code>QueryRouter-AI/ │ ├── ui.py # Streamlit interface ├── Groq.py # LangGraph graph construction and execution ├── Search_fetch.py # Search and retrieval pipeline (SerpAPI) ├── flow.png # Architecture diagram ├── docker-compose.yml # Optional container setup ├── requirement.txt # Python dependencies └── README.md </code></pre> <h2>Technology Stack</h2> <ul> <li>Python 3.10</li> <li>LangGraph</li> <li>LangChain Core 0.2.x</li> <li>Groq LLM API</li> <li>SerpAPI</li> <li>Streamlit</li> <li>FAISS / RAG-ready architecture</li> </ul> <h2>Setup</h2> <pre><code>git clone https://github.com/Rithan377/QueryRouter-AI.git cd QueryRouter-AI python3.10 -m venv venv source venv/bin/activate pip install -r requirement.txt </code></pre> <p>Required API keys:</p> <ul> <li>Groq API Key</li> <li>SerpAPI Key</li> </ul> <h2>Purpose</h2> <p>QueryRouter-AI demonstrates <strong>structured LLM orchestration</strong> where execution paths are explicitly defined through a graph rather than relying on single-call prompt logic. This approach enables easier debugging, modular experimentation, and more production-aligned AI workflows.</p> <h2>Future Improvements</h2> <ul> <li>Advanced routing strategies</li> <li>Memory-aware query handling</li> <li>Vector database integration</li> <li>API-only (headless) deployment mode</li> <li>Evaluation and observability tooling</li> </ul> <h2>License</h2> <p>MIT License</p> <p><strong>Full Changelog</strong>: https://github.com/Rithan377/QueryRouter-AI/commits/v1.0.0</p> |
| title | Rithan377/QueryRouter-AI: QueryRouter-AI |
| url | https://doi.org/10.5281/zenodo.18983870 |