My Portfolio

The system integrates directly with our kiosk application, ensuring a unified ordering experience whether customers order via mobile or in-store.

Customers can choose between coffee, gelato, and other menu categories, with the ability to search the catalog and filter options based on their preferences. If an item offers multiple sizes or flavors, the system dynamically adjusts the selections, allowing for combinations such as choosing a second gelato flavor when ordering a double scoop.

Additionally, it intelligently notifies customers if an item is only available for dine-in and not for takeout, reducing confusion and improving the ordering experience.

Once customers finalize their selection, they proceed to the cart, where they can enter their name and pickup preference—either curbside pickup or dine-in with a specified table number. The system then processes payments via Stripe API, ensuring secure and reliable transactions.

Upon successful payment, customers are directed to a real-time order status page, where they can track progress as their order moves through the system.

For an enhanced experience, the system includes automated order tracking and real-time updates. When an order is received, it is managed through the order processing system, with notifications sent at each stage.

Additionally, built-in geolocation tracking allows for automatic check-in when the customer is within a certain distance of the gelato shop, notifying the cashier to begin preparing the order. This ensures a smooth and efficient pickup process, improving both customer satisfaction and operational workflow.

This project is an end-to-end insurance document comparison workflow built with React on the front end and Firebase (Auth, Storage, Firestore, Cloud Functions) on the back end. The app is gated behind Firebase Authentication, so only authorized staff can access the tool and work with claim data.

Users start by entering a Claim ID that matches the client’s full name, which is used to namespace all Storage and Firestore paths. For each claim, they upload exactly two PDFs (typically the Funari estimate and the carrier estimate). Files are stored under a structured path like claims/{claimId}/, and an upload counter resets the selected documents to keep the UI aligned with new uploads.

A Cloud Storage trigger automatically runs Google Vision OCR on every uploaded PDF, stitches page text together, and writes the extracted content into Firestore under claims/{claimId}/documents/{fileName}. This transforms static PDFs into searchable, structured text ready for downstream AI analysis.

When two documents are selected in the UI, the app calls a callable Cloud Function compareInsurancePDFs. That function pulls both OCR texts from Firestore, feeds them into a specialized prompt, and sends the request to the Gemini API via a dedicated comparison helper. The model returns a GitHub-flavored Markdown synopsis containing an overall summary table, per-area breakdowns, and detailed line-item differences with real currency and percentage calculations.

On the front end, the Markdown is cleaned for valid table syntax and rendered into an interactive result pane. From there, users can:

• Download a PDF of the entire comparison view using html2pdf.js with page-break-aware settings.
• Export Excel workbooks by parsing each Markdown table into individual sheets with cleaned, de-duplicated sheet names.

The system also includes advanced Cloud Functions that parse the AI synopsis, align differences back onto the original carrier PDF coordinates using Vision’s bounding boxes, and draw semi-transparent highlight rectangles over the relevant monetary values. The highlighted PDFs are saved back to Storage and can be surfaced to the UI as signed URLs, giving adjusters a visual overlay of where the carrier is underpaying or out of sync with the public adjuster’s estimate.

Overall, this platform turns a manual, hours-long comparison task into a fast, repeatable workflow with AI-generated summaries, visual highlights, and exportable artifacts for negotiation, documentation, and client reporting.

This project is a fully client-side 3D application rendered with WebGL via Three.js. The scene is composed using hierarchical object grouping to manage planetary orbits, rotations, and relative positioning efficiently.

Each planet is created as an independent mesh with its own orbital group, rotation logic, and invisible hitbox to improve click detection. Raycasting is used to detect user interactions, triggering smooth camera transitions and contextual information panels.

Orbital motion is driven by time-based animation using delta time calculations, ensuring consistent behavior across different devices and frame rates. Visible orbit paths are generated procedurally using ellipse curves for clarity without overwhelming the scene visually.

The asteroid belt and Kuiper Belt are generated using buffer geometries and point clouds for performance efficiency, allowing thousands of objects to be rendered with minimal overhead. Custom collision-avoidance logic ensures that procedurally generated comets do not intersect with planets or the Sun.

Lighting is handled through a combination of ambient and point lights, enhanced with additive sprite-based glow effects to simulate solar radiation. Camera controls use damping and interpolation to provide smooth navigation without sacrificing responsiveness.

The project is deployed as a static site using Firebase Hosting, with all logic executed client-side. The code is structured to prioritize clarity, predictable behavior, and long-term maintainability.

The dashboard provides a high-level overview through dynamic summary cards, real-time clock and weather data, service health indicators, and curated news feeds. It is designed to feel like a live operations or intelligence panel, presenting complex information in a clean, readable format.

Under the hood, the application uses a feature-based architecture with TanStack Query for data fetching, caching, and background updates. Domain-specific hooks abstract API logic, allowing the UI to remain responsive while simulating a realistic production environment.

The platform includes dedicated sections for market tracking (stocks and cryptocurrency), security insights such as CVE monitoring, service outage awareness, and upcoming media releases. System settings allow users to toggle compact mode, manage local cache state, and view environment and build information.

The entire UI is built with modern, framework-free CSS using design tokens and a dark theme system. Firebase Hosting is used for deployment with SPA-safe routing, ensuring smooth navigation and reload behavior across all pages.

Business hours are seamlessly integrated into the system, ensuring that when the store is closed, the display automatically switches to a "We Are Closed" page and returns to the main menu when open. This eliminates the need for manual updates and ensures customers always see the right information at the right time.

All videos and animations can be changed remotely, allowing businesses to swap out promotional content, seasonal specials, or featured items instantly. The menu itself is fully interactive, enabling real-time updates to flavors, descriptions, images, and pricing.

The system also includes inventory control, allowing users to mark flavors as available or out of stock with a simple update from the admin panel. This ensures that customers never see outdated items or unavailable options, improving their overall experience.

With a secure and easy-to-use admin panel, all aspects of the digital signage can be managed from anywhere, eliminating the need to be physically present to make updates.

Whether it's adjusting business hours, modifying promotional content, or updating menu availability, everything is controlled effortlessly from a single interface. This digital signage solution streamlines operations, enhances customer engagement, and provides a modern, visually appealing way to present dynamic content in any business setting.

Upon interaction, users can browse through the menu, view detailed descriptions of each item, and make informed selections. The interface is user-friendly and designed for quick, easy navigation.

All menu items, including images, descriptions, and availability status, can be updated in real-time through a dedicated admin panel, allowing for effortless content management.

If an item is sold out or temporarily unavailable, the system reflects these changes instantly, ensuring customers always see accurate and up-to-date information. This helps streamline operations while enhancing customer satisfaction.

The kiosk application is ideal for reducing wait times and improving customer engagement by offering a self-service experience. Whether it's browsing flavors, exploring new product offerings, or learning about seasonal specials, customers can interact with the system at their own pace.

This solution is fully adaptable to business needs, providing a modern, interactive way to present menu options and promotional content while maintaining full control through a remote admin interface.

This system ensures accurate, real-time inventory updates while simplifying the workflow for store employees and management.

The system features a dynamic catalog view that provides instant access to all inventory items, complete with auto-generated UPC barcodes for seamless scanning and tracking. Employees can quickly search, filter, and sort items based on various parameters, significantly reducing the time spent managing stock levels.

A robust inventory tracking module enables employees to update stock quantities, flag items for review, and monitor stock movement in real-time. Inventory status is categorized efficiently, allowing employees to distinguish between available, low-stock, and out-of-stock items at a glance.

To enhance decision-making, the system incorporates advanced data visualization, providing management with real-time analytics, trends, and statistical reports. The dashboard includes interactive graphs and summary metrics, offering a comprehensive overview of stock levels and sales performance.

A user-friendly interactive table allows authorized employees to modify inventory details, update quantities, and attach additional information. Changes are saved instantly to Firestore, ensuring real-time synchronization across all stores.

Security is a core focus, with password-protected access ensuring that only authorized personnel can manage inventory data. Role-based authentication prevents unauthorized modifications while maintaining a secure and controlled environment for inventory management.

Additional features include real-time stock updates, seamless data synchronization, intuitive UI design, and extensive filtering options to refine inventory searches. The system is optimized for speed and reliability, ensuring a smooth experience for employees handling daily stock operations.

The system features an intuitive dashboard that dynamically updates tasks based on their status. Users can create tasks with detailed descriptions, set priority levels, and specify due dates and times. Tasks are automatically categorized into active and completed sections, ensuring a clear workflow. The system allows users to update task statuses, edit details, and delete tasks as needed.

Key Features

The system is built with React and React Router for seamless navigation and state management. Firebase Authentication ensures secure login and user-based task access. The Firestore database provides real-time syncing and cloud storage for tasks.

Tasks are categorized into pending, in progress, and completed sections for better organization. The completed tasks section displays only the last three items to prevent clutter. A dark mode toggle is included with state persistence using localStorage.

The UI is interactive, dynamically updating using React hooks like useState and useEffect. This ensures smooth transitions and efficient data handling.

Technical Implementation

The Task Management System is built with React functional components, utilizing React hooks for state management. The dashboard is dynamically rendered using the useState hook, while the useEffect hook ensures real-time data fetching and synchronization with Firestore. Task operations such as creation, updates, and deletion are handled through Firebase Firestore, providing seamless cloud storage and access control.

React Router is used for client-side navigation, enabling smooth transitions between pages such as task details, dashboard, and authentication screens. The dark mode toggle is implemented with state persistence via localStorage, ensuring that user preferences remain consistent across sessions.

The application is optimized for performance, leveraging Firestore indexing for fast data retrieval and ensuring minimal re-renders using efficient state management techniques. The UI follows modern design principles with CSS-based styling, offering a clean and professional appearance.

Access to the Task Management System is password protected, requiring users to authenticate before managing tasks. This ensures secure access to task data while allowing users to manage their workflow from any device.

The portfolio includes a variety of projects, ranging from custom WordPress builds and API integrations to full-stack web applications using modern frameworks. Each project is accompanied by detailed descriptions, live demos, and direct links to repositories where applicable. The site is designed for easy navigation, allowing visitors to explore different categories of work and gain insight into my development process. CommitStack serves as an evolving portfolio, regularly updated with new projects and enhancements to demonstrate my expertise in software engineering.

Beyond simple code analysis, AI Code Analyzer serves as a virtual coding assistant by offering real-time suggestions for refactoring and optimizing code. It helps detect redundant logic and improve efficiency while identifying potential issues and suggesting fixes for debugging.

The system also predicts how a codebase might evolve based on the latest industry standards, ensuring that developers stay ahead of best practices and maintain clean, maintainable code.

Additionally, the platform provides an interactive final code preview, allowing developers to visualize how their modified code will function before implementation.

With real-time synchronization, staff can track incoming orders, update order statuses, and process transactions without delays.

When an order is placed through the mobile app or kiosk, it is instantly added to the order queue, allowing staff to view customer details, item selections, and special requests. The system provides a clear visual representation of active, in-progress, and completed orders, reducing errors and improving workflow. Orders can be marked as in preparation, ready for pickup, or completed, with automatic notifications sent to customers through the mobile app to keep them informed in real time.

The Order Manager also incorporates geolocation-based automatic check-ins, allowing staff to be alerted when a customer is approaching for pickup. This ensures that orders are prepared just in time for arrival, minimizing wait times and improving the overall experience. The system is designed with an intuitive interface, making it easy for staff to update orders, handle modifications, and manage inventory availability directly from the admin panel.

By providing seamless integration, real-time updates, and automated customer notifications, the Order Manager enhances the efficiency of handling high volumes of orders while maintaining a smooth and reliable operation.