autonomous hr chatbot

Report Abuse

Basic Information

This repository provides a prototype enterprise application: an autonomous HR chatbot designed to answer HR queries by combining retrieval, tool use, and LLM reasoning. It uses LangChain agents and tools to orchestrate model calls and programmatic actions, Pinecone as a vector database for storing document embeddings, and OpenAI ChatGPT or gpt-3.5-turbo as the LLM. The front end is a Streamlit app that uses the streamlit_chat component. The code includes examples and instructions to store embeddings, run a local or Azure-backed backend (hr_agent_backend_local.py and hr_agent_backend_azure.py), and launch the UI via a Streamlit command. The repo is positioned as a demonstration of how to wire document retrieval, data manipulation, and math tools into an interactive HR assistant rather than a production-ready product.

Links

App Details

Features
LangChain agents and tools orchestrate calls to the LLM and external utilities. Pinecone is used to store embeddings for HR policy documents created with OpenAI"s text-embedding-ada-002 model. A sample Timekeeping Policies document is embedded and searchable for retrieval-augmented answers. Employee data is provided as a CSV and exposed to the agent via a pandas dataframe manipulated through LangChain"s PythonAstREPLTool. Numeric computations are handled by LangChain"s LLMMathChain calculator. A Streamlit front-end using the streamlit_chat component provides an interactive chat UI. The repo includes a notebook to store embeddings and supports an Azure OpenAI variant via a separate backend file.
Use Cases
The project demonstrates how to build an autonomous HR assistant that can answer policy questions, fetch or compute employee-specific details, and combine multiple tools in a single conversational flow. By storing policy text as embeddings in Pinecone, the agent can ground responses in specific documents rather than hallucinating. The ability to manipulate an employee CSV via a Python REPL tool lets the agent run queries, aggregate leave balances, and return data-driven answers. The Streamlit UI enables rapid interactive testing and demos. Instructions and example files make it straightforward to run locally or adapt to Azure, helping teams prototype enterprise HR automation and proof-of-concept integrations with minimal setup.

Please fill the required fields*