Selected work across embedded systems, automation tooling, and full-stack development.
An OCR-powered automation tool that indexes and filters Twitter/X screenshots based on extracted timestamps. Uses PaddleOCR and OpenCV to detect and parse date/time, with exact, windowed, and time-range matching. Includes a Streamlit drag-and-drop UI and ZIP export for bulk results. Containerized with Docker and deployed on Railway for production-ready access.
A smart hiking system combining embedded hardware and mobile software to deliver real-time environmental and motion data. A Raspberry Pi–based vest collects sensor data and communicates with a cross-platform app using BLE and MQTT. Supports features like step tracking and safety-related event detection. Focused on reliable sensor integration, efficient wireless communication, and clear real-time presentation.
An embedded security system built on STM32 using C and FreeRTOS for responsive real-time behavior. Integrates keypad input, OLED + LED status, motion detection, and buzzer alerts. Uses multiple tasks to handle input, sensor monitoring, and feedback concurrently. Emphasized task coordination and reliable hardware–software interaction.
A full-stack task manager built with React, Node.js/Express, and MongoDB with JWT authentication. Designed RESTful APIs to create, update, and delete tasks while keeping data flow consistent between client and server. Implemented a minimalist UI focused on clarity and speed. Applied Git-based workflow, error handling, and scalable backend patterns.