PhD Student · UCSD CSE
Press Enter to open portfolio
I am a Ph.D. student in Computer Science and Engineering at UC San Diego. My work sits around systems, compilers, GPU code generation, and practical AI tooling. I like building things that are technically grounded, useful in practice, and presented clearly enough that the engineering signal is easy to evaluate.
You found the developer's secret room!
Click to open
LEGENDARY REWARD
+500 score, +100 progress, full health/focus restore.
Academic path and current research home, presented as a concise timeline from Tsinghua University to the Ph.D. program at UC San Diego CSE.
Languages and tools I use most often, with a quick visual summary of working strength and day-to-day usage across systems, backend, and research workflows.
Selected research output, highlighting paper titles, venues, collaborators, and the technical problems each project set out to address.
Industry and research experience across labs and companies, focused on scope, ownership, and concrete technical outcomes.
Selected projects spanning operating systems, real-time communication, evaluation infrastructure, and LLM systems work.
Implemented the networking component for a monolithic kernel OS and treated the project as a systems-oriented graduation deliverable.
Built a real-time chat system around Django and WebSocket flows, with emphasis on reliable message delivery and practical web integration.
Built a grading and evaluation workflow for coursework submissions, using Rust to keep the platform lightweight and systems-oriented.
Frames GPU code generation as a measurable agentic workflow problem, with benchmarks meant to surface systems behavior rather than prompt-only demos.
The portfolio opens into a playable systems layer here. Research sections stay primary above, but below this point the site behaves more like a small world with persistent state, companion management, and encounter loops.
The interaction layer is meant to feel like a compact adventure system rather than a disconnected widget. Encounters, capture odds, companion progression, and traversal upgrades all feed into the same persistent loop.
Notes: the grass patch triggers encounters, lower target health improves capture odds, and some companions unlock faster traversal modes.
World navigation for the interactive layer, connecting the portfolio sections with the companion, travel, and challenge systems below.
The anchor point: identity, advisor, affiliation, and direct contact.
Publications and research milestones framed as major unlocks.
Projects positioned as artifacts, systems, and engineering outcomes.
Capture, party management, dex progression, and mount systems.
A designed progression path with a field station, encounter gates, and a route that feels more like a small campaign than a set of loose buttons.
External links stay practical here: GitHub, LinkedIn, and Scholar remain direct exit points for portfolio evaluation.
A challenge route for the interactive layer: recover resources, unlock travel, and push through staged rival encounters with persistent progression.
The field station anchors the route. Restore the active roster, refill core items, and trigger evolution checks before moving deeper into the challenge ladder.
An analytic rival built around control and sustain. Clearing Mika improves resource flow and opens the next layer of route progression.
A more aggressive rival with fire-and-dragon pressure. Tao functions as the route gate, pushing the companion and item systems to matter together.
Press SPACE to test the combat loop.
Type this code as fast as you can!
You were defeated!
Match the programming language pairs!