Junior Fall in Review

Overall: a semester that clearly got away from me#

Where to start.

This was without question my worst semester since starting college. Or more precisely: from the summer internship onward, everything slowly drifted off its original track and out of my control.

A stack of bad decisions, with life, coursework, research, and internship hunting all blowing up in the same semester. A few times I came close to breaking down. It wasn’t any one thing — it was every problem showing up at once, a systemic loss of control.

Living situation: a bad call#

It probably starts back in February. I had lined up an internship and couldn’t be bothered to deal with moving, so I renewed the apartment. I hadn’t realized how enormously important picking the right roommates is, and how directly it determines an entire year’s quality of life.

My requirements for roommates are low: just don’t disrupt my life. So I casually agreed to share with two Chinese guys and one existing roommate. In the end the Chinese guys backed out and became unreachable, and as luck would have it the apartment assigned me two Indian roommates instead. That was the start of the nightmare.

The kitchen and common areas were naturally taken over by the two of them. The kitchen got so dirty and smelled so bad it was unusable, and coming home from campus to that smell was nauseating. What I couldn’t stand most was the two of them shouting in the living room every day — noise-cancelling headphones didn’t help, I had to wear earplugs. It seriously disrupted my ability to study and live there.

Coursework: high-risk, low-return course selection#

I also took several courses that drained an enormous amount of energy. Nonlinear optimization ate huge amounts of time every week and the midterm still crushed me, so I dropped it. That may be the one unambiguously correct decision I made this semester — cutting losses in time.

The other punishing course was operations research, again with an Indian professor who I felt was far too careless about it: poor teaching quality, an accent I couldn’t follow, and no detailed syllabus or explanation for the exams or grade distribution. I knew nothing until the grades were posted, and all I got was an inexplicable GPA — the only course since coming to the US that wasn’t a 4.0. I still don’t know what happened. A big part of it was also being dragged down by a teammate who did essentially nothing on the final project, leaving me to solo it. Looking back, I could have picked better courses. I assumed that taking hard courses meant learning more, when in reality they burned a lot of energy without the corresponding return — I’d have been better off studying the material on my own. Lesson thoroughly learned. I am here at university for two things: a diploma and a good transcript. If I want to learn something I can just learn it myself — CS learning resources are abundant, and there’s AI on top of that. There is no need to make life hard for myself with thankless courses.

Research and internships: the main quest got squeezed out#

Because coursework and living conditions ate so much energy, research and internship hunting — which should have been the main quest — got shelved instead. I found an advisor I liked in the first few days of the semester and got a topic I’m interested in, RL for LLMs, but I was ground down by one thing after another and never put enough time or energy into the research. I’m still in the Slack workspace, but I’ve had nothing substantive to show for a long time, and I feel very guilty about it.

Internship hunting went even worse. I bombed OAs, which are normally my strength — not because I couldn’t do them, but because I kept running into problems and could never get into a good state for them. Interviews went the same way. Nothing at all from big tech, and in the end I took the return offer from the summer and stayed on.

Some thoughts from the internship hunt#

The market is bad, and a lot of the glossy profiles on LinkedIn belong to people hyping and outright lying — turning a provincial second prize in high school into a national first, hearing a term once and immediately calling themselves an expert in it, fabricating things without blinking. And the current environment really does reward the people who perform and talk well. The polite word is “packaging”; the blunt word is fabrication.

In the US most people are willing to trust each other: you don’t have to produce proof of an internship, and nobody asks to see award certificates. But exploiting that trust to lie for personal gain is something I can’t do, and I have no respect for people who do. I don’t think my standards are unusually high — I think theirs are unusually low. To me this is the most basic level of personal integrity.

The result#

That’s the main picture: life, coursework, research, internships — failure across the board.

Grades came out yesterday and I’ve officially said goodbye to the 4.0. Cumulative GPA is 3.978. My remaining courses are basically easy A’s, so I should graduate around 3.98+. GPA doesn’t matter that much, but losing it without ever understanding why still stings.

It’s been a week since the semester ended. I played games for days on end and slept until afternoon. Either way the semester is over and can’t be changed; what matters is what I do next.

Practical analysis and adjustments#

Let me calmly work through how to fix the problems above.

First, the living situation has to improve. I badly underestimated how much sustained noise and disgust degrade cognitive capacity. Uncontrollable, continuous noise and a filthy living room passively drain your energy every day and wreck your mood — you simply can’t focus or stay calm in that environment. I do have a well-paid part-time offer next semester, so in theory I could move. But first I’ll try adjusting my schedule, talking to the roommates and at least getting them to stop disrupting other people, and staying on campus during the day. If that fails, I’ll ask the apartment for a transfer or move somewhere else.

Second, course selection. This semester taught me that difficulty does not equal learning efficiency or return, and that vague grading plus an opaque syllabus equals high risk and low return. I won’t take that kind of unnecessary risk again.

Third, identifying and returning to the main quest. I felt guilty about research and internships many times this semester, which is telling in itself — it means I can recognize what actually matters. The problem was that the two issues above consumed so much energy that I had none left for the things that count. Research doesn’t need a heroic catch-up, but over this winter break I at least need to learn enough of the background, largely finish the project my advisor gave me, and have something to show him before break ends. ICPC NAC is also less than three months away. This is the most realistic shot at the World Finals we’ve had — my teammates are excellent — and since starting college I have never put enough time into getting my competitive form back. I would deeply regret missing this chance. Whatever happens, I don’t want to leave too much on the table. Winter break rehab training plan, let’s go.

Finally, some reflection on the “fabrication and packaging” question. I asked my cousin, who has been working for years, and he made an interesting point: companies and candidates are in a two-way selection process, both testing and misleading each other. Plenty of companies advertise benefits and compensation that aren’t real either — that’s deceiving candidates too. My frustration really comes from equating “I refuse to fabricate” with “I must therefore lose out”, which leaves me both resentful and at a disadvantage. I don’t need to become the kind of person I dislike, who lies as easily as breathing. But I do need to learn to present myself: tell only the truth, but choose the parts of the truth that work in my favor. Change the framing rather than inventing things. That’s a necessary survival skill.

Mindset: reflection and change#

Last, and most important, is the mindset problem.

I have to accept something: I am past the stage of needing to prove I’m not bad. Ever since my losses in competition and the gaokao, I seem to have been chasing perfection — top of the major, a perfect GPA, better than everyone around me. Put plainly, I wanted to prove my worth by winning.

Behind that chase is an implicit proposition: “if I were just a bit stronger, this wouldn’t have happened, I wouldn’t have failed like I did in competition and the gaokao.” Fundamentally I am still trying to correct an earlier failure, to prove that losing back then wasn’t because I wasn’t good enough or smart enough. But the proposition itself doesn’t hold, and chasing it only pulls me in deeper.

There is nothing wrong with pursuing excellence. The real problem is that I picked the wrong, quite toxic, basis for comparison. I set other people as the reference point and built my drive on fear, shame, and a strong urge to prove something. The result is that no matter how well I do it never feels like enough, and I live in a constant low-grade anxiety. The healthier and more sustainable version sets the goal itself as the reference point, with drive built on curiosity, interest, and a sense of mastery.

Objectively, I already have plenty of evidence that I’m “not bad”: several ICPC golds, a high GPA, a return offer from an AI company, a well-paid part-time offer, a clear research direction, and an advisor I like. My parents have reminded me countless times not to over-chase perfection — to get out more, play, rest, relax, be social, widen my life and my experience. And they’re right. Continuing to fixate on proving I’m not bad won’t actually make me feel more secure; it just postpones a harder question: if I’m already good enough, then what am I living for, and what should I be doing?

That question is obviously much harder to answer than “how do I prove I’m stronger”.

So the dangerous thing right now isn’t playing games all day or not working hard enough. It’s continuing to evaluate my future life with the old metrics. I need to break out of the loop and swap in a more reasonable function for evaluating myself. The old system meant that after doing something well I immediately wanted to do it better, so success never brought peace — and one failure led to writing myself off entirely.

From now on, facing a decision, the question can’t be “can I do this perfectly”. It has to be:

• Does this help my larger goal? Nonlinear optimization, for example, was hard and cost me a lot of energy, but it didn’t actually help my research direction or my larger goals.
• Does this build transferable ability that helps elsewhere? Competitive programming, for instance, transfers extremely well — the thinking and reasoning habits I built in secondary school have paid off enormously in college.
• If it does help, can I sustain it, or find a way to sustain it without coming to hate it?

If it broadly satisfies those, it’s a good enough decision. There’s no need to fixate on the optimum.

What I need to learn now is not how to be the best, but how to allocate my energy sensibly — staying physically and mentally healthy while consistently doing the things that serve my ultimate goals.

So, back to the question above: which core capabilities should I build next? Not achievements — mindset.

• First, facing failure calmly. Not simply shrugging off wins and losses, but refusing to escalate a failure into “I’m not good enough as a person”. Put failure back where it belongs: it’s an assessment of one thing, not a verdict on me. After the CSAPP midterm, for instance, I realized I’d done badly, left dinner with friends halfway through because I felt ill, and shut myself in for days convinced I wasn’t capable. (I ended up with full marks in that course, bruh.)
• After a setback, analyze the cause calmly and prepare for the next attempt. Failure doesn’t take away my right to place another bet. Writing this review is exactly that.
• Cut losses early; don’t sink into sunk costs. I realized within the first few weeks that nonlinear optimization wasn’t what I expected, but because I’d already put energy in I kept trying to force it, and dropped it anyway in the end. Rationally I knew I should drop it; emotionally I wouldn’t let go of the sunk cost. Next time I see that a path’s ROI is wrong, I need to decide decisively.
• Resilience. Allow myself to break down, admit it, then analyze and repair afterwards and get back on track. That is far more rational than blindly forcing my way through.

I don’t expect to instantly become someone emotionally steady, mature, and permanently calm. But I hope that after this semester, and after writing this, I’ve moved a little closer to the version of myself I want to be.