NL2SQL CrewAI Analyst
Python
Streamlit
CrewAI
MySQL
Generative AI
May 2026
NL2SQL CrewAI Analyst is a Streamlit app that lets anyone query a MySQL database using plain English — no SQL knowledge required. An AI agent built with CrewAI and NL2SQLTool interprets the question, generates a SELECT query, runs it, and returns both the result and a human-readable explanation in seconds.
The app features a configurable sidebar for entering database credentials and an editable schema pane so the agent always understands your table structure. All previous Q&A sessions are preserved as expandable history cards, and a Clear Session button resets the workspace instantly.
This tool is ideal for analysts, product managers, or any non-technical stakeholder who needs fast self-serve access to database insights without writing SQL — making data democratisation practical and secure by restricting the agent to read-only SELECT queries.
GitHub:
NL2SQL CrewAI Analyst Repository
SQL Chatbot using Crew AI
Crew AI
AI Agents
RAG
LLM
Streamlit
May 2026
This project is an AI-powered SQL search assistant built using the CrewAI framework, designed to translate natural language queries into actionable database operations. It demonstrates how multiple AI agents collaborate to understand user intent, generate SQL queries, execute them on a connected database, and return meaningful insights. The system leverages LLM capabilities along with tools like Ollama or Groq to enhance performance and responsiveness.
The project includes structured workflows where agents are assigned specific roles such as query generation, validation, and execution, ensuring reliable and accurate results. It also integrates sample datasets and databases to simulate real-world analytical scenarios, making it a practical demonstration of AI-driven data querying.
Overall, this solution showcases how agent-based architectures can simplify database interactions, enabling non-technical users to query data conversationally while maintaining scalability and efficiency in backend processing.
Video Link:
Chatbot demo video
Data Processing Using Streamlit
Python
Data Cleaning
LLM
Streamlit
APR 2026
This project is an end-to-end interactive data analysis and preprocessing application built using Streamlit. It allows users to upload raw datasets (CSV/Excel) and perform structured data cleaning, transformation, and exploratory analysis through an intuitive UI.
The app provides modular preprocessing capabilities including missing value analysis and imputation, outlier detection and treatment, symbol and text cleaning, categorical encoding, and feature scaling. Each step is interactive, giving users control over column selection and transformation methods, along with real-time data previews and visual insights.
It also integrates an LLM (via Groq) to generate automated data summaries, highlighting trends, data quality issues, and actionable recommendations—bringing intelligence into the preprocessing workflow.
A key feature of the application is its ability to export preprocessing steps as reusable artifacts (pickle files), such as encoders, scalers, imputers, and outlier bounds. This enables seamless reuse of the same pipeline on new datasets, making it suitable for production-level workflows.
Overall, this tool bridges the gap between manual data cleaning and automated machine learning pipelines by combining user-driven control, visualization, and AI-powered insights in a single interface.
Video Link:
Streamlit app demo video
PDF data Extraction using RAG and LLM
Python
RAG
LLM
GORQ
APR 2026
This project demonstrates how to generate contextual responses using a Retrieval Augmented Generation (RAG) pipeline built with LLMs. RAG systems extend the knowledge of language models by retrieving relevant information from external documents before generating answers.
Repo Link:
RAG Context Generation- Repo
ACES and PIES XML Conversion
Python
XML Parsing
Streamlit
APR 2026
Developed a set of Python scripts and notebooks that extract, transform, and structure large XML datasets used in automotive aftermarket cataloging standards such as ACES (Application Cataloging Standard) and PIES (Product Information Exchange Standard). The solution automates the extraction of hierarchical XML attributes and converts them into tabular formats suitable for Excel-based analysis and downstream catalog management workflows.
App Link:
ACES Conversion- App
PIES Conversion- App
Repo Link:
XML_Converter- Repo
Company Bankruptcy Prediction
Python
Pandas
Model Pipeline
Data Processing
Model Evaluation
Streamlit
PCA
APR 2026
A Machine Learning project that predicts whether a company is likely to go bankrupt based on financial indicators.
This project demonstrates an end-to-end machine learning workflow, including data preprocessing, dimensionality reduction using Principal Component Analysis (PCA), model training using Support Vector Classifier (SVC), and deployment through Streamlit.
Project Link:
Company Bankruptcy Prediction – App
Repo Link:
Company Bankruptcy Prediction – Repo
Telecom Customer Churn Prediction
Python
Pandas
Model Pipeline
Data Processing
Model Evaluation
Streamlit
SVC
MAR 2026
Developed an interactive machine learning web application that predicts the likelihood of a telecom customer churning based on service usage, contract details, and billing information. The application allows users to input customer attributes such as tenure, internet service type, contract type, payment method, and monthly charges to generate real-time churn predictions. The model was trained using supervised learning techniques and deployed through a Streamlit interface, demonstrating the end-to-end machine learning workflow including data preprocessing, model training, evaluation, model serialization, and live prediction deployment.
Project Link:
Customer Churn – App
Repo Link:
Customer Churn – Repo
Automated Comment Sentiment Analysis using NLP
Python
Streamlit
Transformers
NLP
Pandas
BERTweet
MAR 2026
Developed a Streamlit-based NLP application that performs automated sentiment analysis on user comments from CSV, Excel, or JSON files using a pre-trained BERTweet model. The tool processes large text datasets, handles long comments through chunking, and outputs sentiment labels and confidence scores with downloadable results for further analysis.
Project Link:
Sentiment Checking – App
Repo Link:
Sentiment Checking – Repo
Kidney Disease Prediction
Machine Learning
Data Cleaning
Streamlit
Model Deployment
Feb 2026
This project predicts whether a patient is likely to have kidney disease based on clinical parameters.
It demonstrates a complete end-to-end machine learning pipeline including preprocessing, feature engineering, model training, and deployment.
Project Link:
Kidney Disease Prediction – App
Repo Link:
Kidney Disease Prediction – Repo
Warranty Cost Optimization Dashboard (Power BI)
Power BI
DAX
Data Visualization
Business Analytics
APR 2025
Designed a Power BI dashboard analyzing warranty cost-to-sales relationships using KPI targets, geographic insights, and interactive slicers to support cost optimization decisions.
Project Link:
Warranty Analysis Report
Adventure Works Sales Analytics Dashboard (Power BI)
Power BI
DAX
Data Visualization
Business Analytics
APR 2024
Developed an interactive Power BI sales analytics dashboard for Adventure Works, visualizing sales performance, profit percentage, and product category trends across regions to enable data-driven business insights.
Project Link:
Adventure Works Report
