AVAILABLE FOR SOFTWARE ENGINEERING OPPORTUNITIES
I'm Charles Dai, a Carnegie Mellon University student building usable full-stack products, AI-assisted workflows, and engineering systems where the details matter.
01 / SELECTED WORK
A curated collection of shipped interfaces, research tooling and experiments. Every project links to its actual source; live demos are identified separately.
Understand the impact before a schema change breaks the things downstream.
A DataHub-context review workflow that traces ownership and lineage, analyzes potential blast radius, and drafts safer migration guidance. Focused on evidence-backed decisions rather than opaque AI output.
A cinematic frontend with an engineering system behind the creative direction.
Designed and built a motion-first studio website with Next.js, TypeScript and Tailwind. The repository also includes a Python outreach workflow with deterministic safety checks and human approval boundaries.
AI proposes. Deterministic code decides what is safe to believe.
An architecture for catalyst-driven market research with policy gates, provenance, reproducible outputs and explicit failure states. The public repository presents a curated research layer; production trading is not enabled.
Local narration and accessible visual-cover generation with SHA-256 provenance, and optional Backblaze B2 storage integration.
A deterministic daily puzzle about dependency graphs, blast radius and optimal containment. Built for Reddit's Devvit environment.
A classroom budgeting app with a Node.js API, server-computed integer-cent totals, validation and automated tests. Fictional data only.
A deliberately simple QR generator: one purpose, a focused interface, no account and no application backend.
Explore the assumptions behind Anthropic and OpenAI valuations.
An AI-assisted independent project by Charles Dai, pairing a reviewed economic cash flow model with dated research, a shared browser/Node.js engine, and Excel/PDF evidence. Model and API tests cover the cases where a plausible result can still be wrong.
PV of forecast
+ PV of terminal value
02 / ENGINEERING APPROACH
Make the purpose clear. Design the interaction around an actual task. Reduce complexity at the surface without hiding it under the hood.
Prefer reproducible tests, source links and explicit constraints. A meaningful failure report is more useful than an unearned green checkmark.
Improve the visuals, the system boundaries and the developer experience together. Document what changed and why it matters.
03 / ABOUT
My interests live at the intersection of software engineering, product design and intelligent tools. At Carnegie Mellon University, I explore how emerging AI capabilities can translate into practical, human-centered experiences.
I enjoy working across the stack: a thoughtful frontend, an understandable API, and the underlying tests and operational safeguards that make the whole thing easier to trust.
Explore my code on GitHubPython · TypeScript · JavaScript · SQL · HTML/CSS
React · Next.js · Tailwind CSS · Responsive UI · Accessibility
Node.js · REST APIs · Data validation · Agent workflows
Git · GitHub Actions · Automated tests · Playwright · Vercel
04 / SAY HELLO
I'm interested in software engineering, AI engineering and product-focused internship opportunities. If something here resonates, I'd be glad to connect.
Contact routes through my publicly listed studio inbox. No form submissions or tracking are collected on this site.