This post comes rather late — mostly because I procrastinated, partly because I needed distance to understand it all. Looking back, 2025’s application season was a strangely emotional yet enlightening ride.

Background and Outcome

I work on video world models, with previous experience in LLMs, agents, and interpretability. I did my undergrad in Computer Science (IEEE Honor Class) at Shanghai Jiao Tong University and applied to both PhD and Master’s programs.

Results summary:

  • UIUC CS PhD offer
  • UCSD PhD (verbal offer, later declined)
  • Several Master’s offers, including CMU MSML, which I finally chose.

The Early Chaos — Lost in the Game of Grades

In my freshman and sophomore years, I was an ambitious yet clueless player in the system: chasing GPA, joining competitions, juggling meaningless clubs. I thought being “busy” equaled being “excellent.” In truth, I was wandering aimlessly within an echo chamber of mediocrity.

I even signed up for a math double major — thinking it would make me “smarter.” It didn’t. It turns out we should follow Andrej Karpathy’s advice: “learn when you need it.” Courses like Real Analysis and Abstract Algebra were intellectually beautiful but practically detached from my research path. By the time I realized it, I was exhausted, not enlightened.

What I should have done? Talk more to seniors. Information asymmetry kills curiosity faster than ignorance. If I could talk to my freshman self, I’d say: stop optimizing for GPA, start optimizing for perspective.

Build Connections

Every meaningful pivot in my journey started with a conversation. Mentors, friends, even strangers on X (Twitter) — each shared a fragment of the map I was missing.

“Connection is all you need.”

Letters, publications, interviews — all are proxies. But real connection is how opportunities flow in academia. Professors invest years in a PhD student; they don’t gamble on cold emails, they trust voices they know.

Paper ≠ Value

During my UCSD internship, my advisor once told me:

“Papers aren’t that important. Imagine if conferences collapsed tomorrow — would your work still matter?”

I didn’t fully agree back then. But as conference submissions ballooned into tens of thousands, I started to see his point.

To evaluate a candidate’s research ability, a paper’s worth lies in its originality and authorship, not its citation count (if one was just involved, not in a leading role). Independent first-author work > co-first author > “others.”

And for PhD admissions? Return offers and collaborations speak louder than PDF files. It’s not about “who has more papers,” but “who has been trusted to build.”

Over-Planning Is the Enemy of Progress

I’ve tried detailed timelines, fancy Notion dashboards, and to-do lists with color codes. They all collapsed under real life.

Perfectionism is just procrastination in disguise. A lesson I learned painfully: during ICLR prep, I wasted days formatting a huge metrics table until my collaborator bluntly said, “We only need five of these. The rest go on the website.” He was right.

Focus on leverage — what actually moves the needle. Everything else is noise.

Read with a Purpose, Think More

Every morning, I used to open Twitter or arXiv, bookmarking papers like a ritual. Most stayed unread.

A Harvard professor once told me:

“Reading too many papers limits your originality. Read the classics. Then explore your own questions.”

Now I follow a simple rule: Start from a few seed papers, expand only when curiosity demands it.

Also — write. Research blogs are powerful if you treat them not as notes, but as teaching tools. If your blog doesn’t express an original thought, it’s just an annotated syllabus.

Why I Turned Down a PhD Offer

This is the question I’ve been asked the most.

Rejecting a funded PhD sounds insane in today’s hyper-competitive environment. But I wasn’t ready to commit five years of my life to something I wasn’t certain I’d love.

A PhD is not just a degree — it’s a lifestyle. If you don’t truly enjoy the process of research, no fellowship or ranking will save you from burnout.

After declining, I did another research internship abroad. It broadened my horizons and gave me a deeper sense of why I wanted to do research in the first place. Looking back, that detour was the best form of progress.

Learning to Embrace Uncertainty

When faced with major life decisions, I classify them into three types:

  1. Safe choices — low risk, predictable path.
  2. Bad choices — obviously wrong.
  3. Uncertain choices — unclear outcome, unclear reward.

Every transformative decision I’ve made falls into the third type: dropping the double major, studying abroad, rejecting a PhD offer.

Life isn’t convex — you can’t find a global optimum without exploring uncertain valleys. A little randomness is essential to escape local minima.

Language and Interviews

For international students: prepare TOEFL early, just to get it out of the way.

Interviews mostly revolve around your research. Most professors expect English slides and clear communication. Rarely, some may ask math or coding questions — always be prepared.

Treat it as a discussion, not an interrogation. Speak with clarity, not perfection. The goal is to communicate ideas, not accents.

Closing Thoughts — Progress, Not Perfection

A senior once told me:

“Doing a PhD should be a lifestyle, not a race.”

That line stayed with me. I’m still learning, still far from perfect. But every choice — right or wrong — taught me something real.

To everyone who helped, encouraged, or just talked with me through this journey: thank you, sincerely.

Onward to the future. 🚀

Helpful Link: