Degree

Master of Science in Computer Science

Department

Department of Computer Science

School

School of Mathematics and Computer Science (SMCS)

Date of Submission

Fall 2025

Supervisor

Ms. Tasbiha Fatima, Lecturer, Department of Computer Science

Keywords

Conversational Analytics; Text-to-SQL; Large Language Models; Self-Service Analytics; Business Intelligence; Natural Language Processing; ClickHouse; Progressive Web Application

Abstract

This report presents the implementation of the Conversational Analytics Platform (Data Chat AI), an AI web application built to democratize data analysis in the enterprise through natural language interfaces. The platform addresses a critical issue: technical barriers prevent about 70 percent of business users from accessing databases independently, creating bottlenecks in data-driven decision making. The final implementation is a production system deployed in a live environment, including: (1) a full-fledged web application with offline capabilities through Service Workers, (2) natural language to SQL conversion with multiple LLMs (GPT-4.1, GPT-5.1, Claude Sonnet 4.5, Gemini 2.5 Pro) through the OpenRouter API, (3) ClickHouse integration for fast analytical query execution, (4) a multi-step research workflow with background job processing, and (5) comprehensive user activity analytics. The backend is built on Laravel 12 and PHP 8.2, with Alpine.js 3.4 and Tailwind CSS for a reactive frontend, ClickHouse for OLAP workloads, MySQL for transactional data, and Redis for caching. Testing is implemented with Pest PHP 3.8, covering authentication, core functionality and profile management. Key outcomes include a deployment serving diverse users, implementation of all 11 functional requirements, and cross-platform accessibility via PWA and Android TWA. The platform demonstrates the feasibility of combining state-of-the-art LLMs with high-performance analytical databases to deliver accessible, business-grade analytics.

Document Type

Restricted Access

Submission Type

Research Project

Download

The full text of this document is only accessible to authorized users.

Share

COinS