projects

Ksama Arora

DocuQuery

Tech Stack: LangChain, ChromaDB, Ollama (Mistral Model), PyMuPDF, python-docx, Streamlit

Developed a retrieval-augmented generation system that enables users to upload PDF, DOCX, and TXT files and extract targeted information using natural language queries.
Leveraged LangChain for managing document workflows, ChromaDB for efficient vector embedding storage, and Ollama’s Mistral model for precise query processing.
Integrated advanced vector search algorithms and contextual matching for reliable document analysis and information retrieval.
You can find the repository here: github.com/ksamaarora/DocuQuery-RAG-LangChain-Ollama.

Oncology Breast Scanner

Tech Stack: Python, TensorFlow, Keras, DenseNet201, OpenCV, Albumentations, Google Colab

Developed a deep learning-based breast cancer detection system using DenseNet201 to classify mammogram images into eight diagnostic categories, including benign and malignant density types.
Implemented advanced data preprocessing techniques, such as contrast enhancement and normalization and leveraged augmented datasets for training.
Achieved high validation accuracy by optimizing hyperparameters, automating the end-to-end workflow and validating predictions on unseen data for reliable medical analysis.
You can find the repository here: github.com/ksamaarora/OncologyBreastScanner.

Healthcare Revenue Cycle Management

Tech Stack: Azure Data Factory, Azure Databricks, SQL, ADLS, Key Vault, Delta Lake GitHub

Designed and implemented a robust data pipeline to optimize healthcare revenue cycle management using Azure Data Factory and Azure Databricks. Leveraged Delta Lake for efficient data storage and processing, enabling real-time insights into billing, claims and patient financial data.
Integrated Azure Data Lake Storage (ADLS) with secure data handling using Key Vault, maintaining compliance with healthcare regulations. Utilized SQL to perform data transformations, ensuring actionable insights for decision-making, improving data processing time by 40%.
You can find the repository here: github.com/ksamaarora/healthcare-revenue-cycle-management.

ForensiFeed

Tech Stack: React, Node.js, Vanilla JavaScript, AJAX, Python.

Developed an automated tool for forensic investigation, capable of parsing social media feeds from platforms like Twitter, Facebook, Instagram and Telegram, reducing manual effort by 80%.
Worked on Back-end of the website and helped my teammates with integrating the front-end interface. Created user-friendly web-based interface that allows investigators to input usernames, hashtags or keywords for automated data gathering, improving data retrieval speed by 60%.
Integrated a web application, increasing accessibility for investigators across different platforms.
More information and the report: github.com/ksamaarora/ForensiFeed.

Winning XI

Tech Stack: Python, Web scrapping, Pandas, Power BI Github

Created a Power BI report to identify top 11 players for a T20 cricket team by scraping data from espncricinfo with a Brightdata website tool, cleaning and transforming the data with pandas, and evaluating various player performance metrics.
Used the resulting Power BI dashboard to select players for various categories (openers, middle order/anchors, finishers, all-rounders, specialist fast bowlers) and ultimately choose the top 11 players to play in the match. Selected team using the Power BI dashboard has 90% of chances to win the game.
You can find the repository here: github.com/ksamaarora/T20-worldcup-cricket-data-analysis.

Sorting Visualizer Live

This project is a Sorting Visualizer that demonstrates the inner workings of various sorting algorithms using a visual representation. The aim is to help people understand how different sorting algorithms compare in terms of efficiency, speed, and the steps they take to sort an array of elements.
The visualizer supports several algorithms, including Bubble, Selection, Insertion, Merge, Quick and Heap.
You can find the repository here: github.com/ksamaarora/Sorting-Visualizer.

Tour Booking System

Tech Stack: PHP, MySQL, HTML, CSS, JavaScript.

Designed a user-friendly interface featuring a landing page, an ”About Us” section, a ”Popular Tours” page with recommended options, and an interactive ”Book Now” form, increasing user engagement by 50%.
Improved booking efficiency leading to a 30% increase in successful transactions.
Focused on ensuring smooth user experience and data management through effective integration of front-end and back-end technologies. Implemented a booking form that collects and stores user details, enhancing data management and reliability.
You can find the repository here: github.com/ksamaarora/Tour-Booking-System.

Library Management System

This project is a Library Management System that utilizes the Binary Search Tree (BST) data structure to efficiently manage books and their quantities. The system is designed to be user-friendly, offering two main interfaces for both administrators and users.
You can find the repository here: github.com/ksamaarora/Library-Management-System.