A simple browser extension I built to automatically win a daily quota war for free HKI registrations at my university.
An audio steganography tool that hides secret messages inside songs by simply pasting a YouTube link.
A complete end-to-end ERP application built to strictly adhere to predefined BPMN specifications for a university course.
I built a scan-first WebAR demo because requiring a 200MB app download just to see a product in 3D is a great way to lose a customer.
I built a rigid API-first decision agent returning strict JSON to prioritize safety over unreliable chat interfaces.
Deaf students deserve real-time accuracy without sacrificing privacy to a cloud server, so I built this offline-first iOS captioning app.
Chronic stress spirals quickly. I built this watch-first app to catch the exact moment it starts and force a critical micro-pause.
Standard climate indices wash out local details. I deployed unsupervised ML on raw buoy data to catch hidden temperature anomalies.
This chatbot proves that coordinating heterogeneous services is far harder than building a single isolated model.
Building this reporting platform taught me that the hardest part of software isn't the code, but navigating rigid institutional rules.
I built this corporate site entirely with AI agents, delivered successfully, then quit because I learned following orders isn't for me.
Online learning is broken and lonely, so I teamed up with a driven co-founder to build a network where people actually learn together.
Building a college marketplace taught me how dual PM-Engineer execution and formal ICONIX methodology keep a team aligned and shipping.
Leading this sprint taught me that product management is simply about untangling complex problems so your team can do their best work.
Stepping away from code to lead a healthcare SaaS strategy taught me how cross-functional dynamics actually drive a product's success.
This backend hardening project forced me to treat release reliability and security controls as a single, repeatable delivery system.
Extracting signal from noisy app reviews to demonstrate how structured preprocessing creates actual analytical value.
Moving beyond isolated notebooks to build a TFX pipeline proved that operational reviewability matters as much as the model.
A straightforward comparison of recommendation algorithms that anchors machine learning choices firmly in the context of user discovery.
This project strips e-commerce to its essentials, forcing a deep understanding of the workflow between a storefront and its admin portal.
A disciplined CV exercise that trains a CNN to distinguish animal classes, prioritizing reproducible pipelines over flashy deployments.
Ignoring predictive hype in favor of exploratory data science to surface environmental patterns that matter for public health.
Treating business rules and use cases as a first-class concern makes for a vastly more resilient architecture.
Treating regression not just as math, but as a business tool to reliably translate property attributes into market insights.
A baseline computer-vision project that proves the value of disciplined data augmentation over unnecessary system complexity.
We designed this tool to guide users through the practical trade-offs required to actually change their carbon footprint behavior.
Separating core API requests from heavy background tasks is crucial to prevent the whole system from grinding to a halt.
Strict payload validation and consistent response contracts are the absolute foundation of any reliable backend system.
A purely structural web project that relies on semantic HTML and precise CSS layouts to deliver a clean, readable historical narrative.
A weather app that prioritizes fast access over feature bloat. Built for a national competition. Took third place.
Treating the frontend as a pipeline shows that optimization and bundling are just as critical as the code itself.
A movie search app that proves vanilla JavaScript can ship with a professional Webpack build pipeline, no framework required.