I'm Nathan Lin

SellRaze is an AI-powered e-commerce platform that enables users to instantly list items across multiple marketplaces by simply uploading a photo or barcode. The app automates the entire listing process, generating optimized posts on sites like eBay, Facebook, Depop, and Mercari, allowing sellers to reach a broader audience with ease.

Currently, I am working on transforming the existing mobile app, initially built with React Native and TypeScript, into a fully functional desktop application. This involves adapting mobile-specific functionality and UI elements to fit the larger screen, while maintaining a seamless user experience. The desktop version is developed using TypeScript, React, and Tailwind CSS, and incorporates optimized design patterns and responsive layouts tailored to desktop workflows, allowing for efficient cross-platform synchronization and streamlined management of sales, shipping, and customer interactions.

This is a web-app that allows occupational therapists to analyze and interact with the data coming from sensors within the SmartSeat to monitor patient movements and assess the physical health of older adults.

The project utilizes ReactJS for the frontend, with a Flask API backend. Data management is handled through InfluxDB and real-time updates are managed via HomeAssistant. Graphing capabilities are done though syncing the data to a Grafana dashdoard to analyze data.

As a university student, UniBuzzy is your go-to app to connect with other students on campus, attend exciting events, and discuss the hottest news.

The application was developed using Vue for the frontend and SpringBoot and OracleDB for backend services, while integrating with external APIs for real-time event management. As the front-end development lead, I led the team by managing 50+ weekly merges and organizing SCRUM meetings to ensure smooth integration and timely feature delivery.

This is an AI-driven project for COVID-19 diagnosis using X-ray analysis, achieving 90% accuracy through research and optimization of various deep learning models.

The model leverages a multimodal late fusion approach , combining image data with tabular clinical data to enhance detection accuracy by incorporating diverse data sources. Implemented in Python with deep learning frameworks such as TensorFlow, the model architecture includes EfficientNet, VGG, and ResNet structures. Optimization techniques like Monte Carlo dropout and Bayesian optimization are applied to further improve accuracy and model robustness.

Punchshot Pickleball is a client-requested project managed by GT WebDev that allows pickleball players to join and create teams and compete in local leagues.

The app was built using ReactJS for the frontend, with backend services powered by NodeJS/ExpressJS and data stored in MongoDB. It integrates APIs like SendGrid for email notifications and Google Maps for geolocation.

This is a generative AI-powered application under Technology Liberation Corporation that assesses the sales performance of a company’s sales team.

The frontend is developed using ReactJS, while the backend is integrated using NodeJS. The application also leverages Google Cloud APIs for data analysis, with prompts generated using the GPT API.