ZukBits Workflows Management System is a centralized internal web platform designed to streamline project management, documentation control, approvals, scheduling, reporting, and user coordination within an organization. The system provides role-based dashboards and structured workflows that enhance operational efficiency, accountability, and visibility across teams.
A secure full-stack Flask application for publishing Jupyter and R Markdown projects with admin controls, analytics, and Supabase storage integration.
A full-stack Python machine learning web application that analyzes customer churn and bank marketing datasets, provides predictive modeling, and visualizes performance metrics through an interactive dashboard. The platform integrates trained ML models, live prediction endpoints, and statistical insights into a deployable web interface.
A fully customized school website built for The Go Getter Academy (Kyaani), featuring a dynamic PHP backend, MySQL database integration, and a responsive front-end design optimized for performance and SEO.
A custom-built PHP 8 portfolio website with a secure admin dashboard, PostgreSQL database integration, Supabase storage support, and clean routing architecture designed for scalable deployment.
A responsive multi-page veterinary and pet care website built as a front-end design experiment. The project focuses on layout structure, UI presentation, and category-based pet service pages without backend functionality.
A full-stack Laravel-based web application designed to manage and showcase professional tech services with a secure backend architecture, structured database management, and scalable frontend integration.
Performs a full-scale analysis of Australia’s Private Hospital Insurance (PHI) ecosystem by importing multi-sheet Excel datasets, repairing inconsistent formats, harmonizing variables, constructing age and product-mix cohorts, and analyzing long-term membership trends pre-COVID, during COVID, and post-COVID. Also builds a complete micro-level logistic regression model using demographic and socioeconomic features to estimate odds of PHI uptake, including VIF diagnostics, ROC-AUC evaluation, marginal effects, and synthetic scenario simulations.
Builds a full NLP pipeline that pulls live BBC RSS headlines, cleans and vectorizes text, applies both VADER and SentiWordNet sentiment models, trains Word2Vec embeddings, and generates dashboards (word clouds, stacked sentiment charts, and summary tables) to compare sentiment across World, Business, and Technology news.
Develops an end-to-end decision intelligence system for determining optimal loan-limit increases. The workflow includes data cleaning, customer behavioral segmentation, acceptance-probability modeling with logistic regression, default-risk estimation, Markov-chain state transitions for evolving risk states, expected-profit calculations, portfolio-level linear programming optimization, 12-month Monte-Carlo simulation, economic stress-scenario modeling, and reinforcement learning (Q-learning) to learn data-driven offer policies. The system maximizes profit while respecting exposure, default, and risk constraints.
Loads multi-sheet cleaned Excel datasets for Australian Private Hospital Insurance (PHI), restructures and harmonizes complex quarterly/monthly blocks, constructs national hospital and general-treatment insured-person series, and produces interactive Plotly dashboards for long-term trend analysis, COVID-period comparisons, state-level heatmaps, age-gender cohort analysis, YoY growth, and composition shifts. Also implements full statistical modeling of PHI uptake using micro-level demographic and socioeconomic variables, including logistic regression, marginal effects, MLS (Medicare Levy Surcharge) policy impact estimation, confusion matrices, ROC-AUC evaluation, and scenario-based prediction changes.
Evaluates how effective different security awareness training types are by cleaning assessment and phishing test data, tuning a Random Forest model to predict post-training scores, comparing it with other classifiers, and using rich visual and statistical analyses (correlations, t-tests, ANOVA) to show improvements and differences across departments and training methods.
Loads raw weather data and performs extensive exploratory analysis using both Matplotlib and Plotly, generating time-series, bar charts, scatter plots, box plots, and multi-subplot dashboards to examine temperature, humidity, and pressure across days, months, years, hours, and cities.
Cleans and preprocesses a large GB YouTube trending dataset, extracts temporal insights, performs KMeans clustering on engagement metrics, fits a linear regression model to understand what drives view counts, computes correlations, applies PCA for dimensionality reduction, and visualizes category trends, top channels, disabled features, and time-based publishing patterns.
Analyzes qualitative survey responses and social media posts to uncover brand associations, product perceptions, and novelty cues. The workflow includes text cleaning, tokenization, frequency analysis, thematic classification, word clouds, sentiment scoring using VADER, location-based segmentation, and comparison of novelty/exclusivity keyword usage across customer groups. It also builds respondent-level theme flags and cross-segment insights for marketing strategy.
Converts raw loan datasets from PDF and Excel into structured tables using Tabula, performs data cleaning, standardizes categorical variables, merges datasets, and generates analytical insights on loan amounts, approvals, demographics, property profiles, self-employment effects, and applicant income distribution with supporting visualizations.
Pulls live remote tech job postings from the RemoteOK API, cleans and analyzes skills, companies, locations, and posting trends, and builds rich visualizations and a mini dashboard to explore in-demand skills and market activity.
This project builds a unified retail pricing analytics system that cleans and merges multi-sheet data, engineers key sales features, analyzes patterns like discounts and seasonality, trains OLS and Ridge models to predict order revenue, and exports a Tableau-ready Excel workbook with insights and diagnostics.