Hello,
  Welcome!

Building management teams need immediate access to every fire prevention device's physical and digital address — especially during false alarm events where speed is critical. Previously, this data was trapped inside the fire alarm panel's internal program, requiring on-site access and specialized software to retrieve. The organization needed a standalone report that could be distributed and referenced without touching the panel.

I built the SDU Device List Maker — a tool that transforms raw, unstructured exports from Edwards Technology SDU versions 2,3 and 4.X into clean, comprehensive device location reports. The output is a print-ready PDF compliant with FDNY Rule 907-01. The software processes data through multiple validation phases, including a user review step before final output, ensuring accuracy at every stage.

This project began as a straightforward request: build a barcode-based system to track product sales. I designed and developed the initial inventory platform from the ground up — a locally hosted database with core tracking capabilities. What started as a simple tool quickly proved its value, becoming the backbone of daily warehouse operations across the organization.

Over the years, the system grew organically based on real user feedback and operational needs. It now includes low stock alerts, multi-warehouse support, mobile access, ideal stock levels, committed item tracking, reordering reports, job tracking, purchase orders, invoice attachments, and full export capabilities. Still operational today, the platform demonstrates how purpose-built software can scale alongside a growing business.

Field technicians routinely update fire alarm panel software on-site, then need to return the exported configuration files to the central server. The challenge: unreliable connectivity in the field, risk of human error during uploads, and no automated way to track versions or notify stakeholders. The organization needed a system that could handle all of this reliably — even on a weak mobile connection.

After interviewing field technicians, I discovered that Microsoft Teams and email consistently worked even in low-connectivity environments. I built a Teams and email chatbot using Power Automate that manages the entire file version control workflow — accepting uploads, tracking revisions, preventing duplicates, and sending automated notifications. A connected SharePoint site gives the management team full visibility. The system has been running for over two years with virtually zero maintenance.

Simple Expense Reports is a full-stack application built entirely from scratch — no frameworks, no boilerplate, no shortcuts. I designed every workflow, interaction, and layout around a clear vision: make expense reporting fast, intuitive, and visually engaging. From the warm manila report cards to the ruled-paper line item design, every detail was intentional. The goal was to build a tool that feels polished and purposeful while solving a real operational need.

The app runs on a Node.js/Express backend with MySQL, a vanilla JavaScript single-page frontend, and hand-written CSS — no heavy dependencies or build tools. I implemented PDF generation with PDFKit, configurable SMTP email submission, multi-currency support, file attachments, and an analytics dashboard with category breakdowns and monthly trends. Building a project with full ownership — from database schema to pixel-level styling — and leveraging AI-assisted coding workflows to iterate quickly remains one of the most rewarding aspects of programming.

Modern business operations demand intelligent workflow automation that goes beyond simple task scheduling. I designed and deployed agentic AI orchestration systems using platforms like Power Automate, Make.com, and n8n — embedding AI-driven logic directly into daily operations. These solutions handle everything from document routing and data extraction to complex decision trees, reducing manual effort and eliminating repetitive bottlenecks across entire departments.

I built multiple virtual assistants from the ground up using the OpenAI API, applying careful prompt engineering and LLM evaluation to refine each integration. Each assistant was hand-coded in JavaScript, PHP, and HTML — communicating directly with the API without relying on third-party platforms. This approach ensures full control over the source code, response behavior, and data flow, delivering tailored conversational experiences purpose-built for each use case.

Each deployment follows a proven architecture: Linux virtual machines on Azure, paired with WireGuard for encrypted tunneling and Samba for seamless file sharing. I configured firewall rules, network routing, and access policies from the ground up — building secure, flexible environments that give remote teams the same reliability as on-premise infrastructure.

Beyond networking, I managed the full cloud stack — provisioning VPN connections, configuring SharePoint as a centralized file server, and enforcing strict access controls through Azure Active Directory. The goal was always the same: protect sensitive business data without adding complexity for end users. This integrated approach allowed multiple organizations to retire local servers entirely and operate securely from the cloud.