QHSE Guidebot

AI chatbot using Llama and RAG for efficient QHSE SOP retrieval

QHSELLMRAGLangChainChromaDB

Features

Tech Stack

FlaskLangChainHuggingFaceMySQLGroq APIChromaDBTailwind CSS

Access Links

Source Code

PreviousSmart Meter Forecasting with Prophet Algorithm and PySpark MLlib Pipeline

NextQHSE Object Detection for Workplace Safety Monitoring

QHSE Guidebot | Toriqul Muchlissin

Project Overview

QHSE Guidebot

QHSE Guidebot is a dummy prototype project that simulates an AI chatbot platform for streamlining access to Standard Operating Procedures (SOPs) in the Quality, Health, Safety, and Environment (QHSE) domain.

By leveraging Retrieval-Augmented Generation (RAG) and Llama-based models, the system allows users to ask natural language questions and receive document-grounded answers in real time.

Core Features

Natural Language SOP Querying

Powered by Groq API with the Llama model for fast, accurate response generation
Integrates RAG pipeline for retrieval-augmented answering
Supports intuitive querying in plain language related to QHSE procedures

Document Retrieval System

Document embeddings using sentence-transformers/all-MiniLM-L6-v2 (HuggingFace)
Embedded vectors stored and retrieved using ChromaDB vector store

Web Interface

Web interface provides different capabilities based on user roles:

User Dashboard

Ask questions via chatbot UI

Admin Dashboard

Manage SOP uploads (PDF/doc)
Monitor user registrations and approve access

Key Achievements

Efficient SOP Retrieval: Utilizes Llama + RAG + ChromaDB to allow real-time question answering over company SOPs
Optimized Embedding & Storage: Leverages all-MiniLM-L6-v2 for lightweight yet effective embedding, stored in ChromaDB
Scalable Role System: Supports flexible access control for admins and users, ensuring document security and proper usage
Enhanced Workplace Safety: Minimizes the time required to locate critical QHSE procedures, supporting better decision-making and regulation adherence