Past Projects

This web-based tool aligns catheter pressure and MRI volume data to enable precise PV loop analysis in pediatric cardiology. Built with Nuxt.js and a MATLAB backend, it reduces human error in data synchronization and supports clinical decision-making in congenital heart disease diagnosis.

This tool automates Zigbee application development by generating code from cluster configurations using the ZAP platform and Handlebars templates. It reduces errors, speeds up development, and improves the TI customer experience by removing manual coding barriers.

This project simulates smart city infrastructure management with OpenCSMP and Cisco FND. It enables provisioning and monitoring of virtual devices like traffic lights using IPv6 and CoAP. Despite platform limitations, it lays the groundwork for secure, scalable device lifecycle management in public networks.

This QA solution integrates TestRigor into SmartData Solutions’ SDLC pipeline. It supports natural language-based test writing, reusable automation rules, and robust coverage (92%) for production web apps. The system significantly reduces manual testing while improving resilience to UI changes.

This system automates diagnostics and health checks on Sagemcom’s broadband gateways for AT&T and T-Mobile. It includes memory leak detection via machine learning, automated core dump reports, and WAN status aggregation cutting debug time and enhancing gateway reliability.

To accelerate TinyML on RISC-V, this project optimized TensorFlow Lite Micro kernels using custom vector instructions. The team built a cycle-level analyzer and achieved 9x kernel-level and 4x application-level speedups. The results validate vector processing as a viable and efficient alternative to scalar execution on embedded ML workloads.

L3Harris’ tool captures and visualizes RF calibration data from airborne systems in real time. Using C++ and Java Swing, it supports live data plotting, replay of historical datasets, and scheduled packet transmission. The system improves calibration accuracy and reduces flight-testing costs.

Trademarkly is a SaaS platform that simplifies trademark filing for individuals and small businesses. Built on AWS with AI-driven similarity checks, it automates status tracking, payments, and document workflows. Users benefit from intuitive dashboards, fast onboarding, and legal process transparency.

Humach’s AI Campaign Manager automates email outreach using APIs like Twilio and SendGrid. It personalizes communications using real-time analytics and sentiment-aware automation, boosting lead engagement and conversions. The system replaces manual campaign execution with intelligent, traceable AI-driven processes.

This project delivered a secure, real-time messaging platform integrated into Folio’s cross-platform desktop and mobile ecosystem. Built with .NET WinForms and .NET MAUI, it features AES end-to-end encrypted chat, SignalR backend, Azure Key Vault integration, and a MySQL database. The system enhances collaboration and communication within Folio’s cloud workspace.