My Job Hunt Journey (Part 1): What It Felt Like to Fail Two ByteDance Interviews

[doggie]

Preface

In “A097 Outside A097 Hospital, a Sky Full of Stars”, I mentioned that medical students aren’t limited to hospitals—they can go many other places. Today, I’ll share my experiences interviewing at ByteDance—twice. Both ended in rejection, but they’re still worth sharing.

I won’t preach or lecture. I’ll only tell stories. How much you absorb is up to fate; if you walk away having learned nothing, at least you’ve had a good laugh.

Main Text

Long, Long Ago

Although I firmly chose the 8-year Traditional Chinese Medicine (TCM) program at Beijing University of Chinese Medicine right after the Gaokao—driven by my passion for TCM—I also developed strong interest in computer science during university (and in many other things, most of which I gradually abandoned). I studied quite a bit (see “A038 My Programming Learning Journey and Takeaways” from 2023), and did some hands-on practice: tinkering with Linux servers, building websites, etc.

I found fiddling with computers far more interesting than practicing TCM—but I was a TCM student, destined to become a TCM practitioner. At that time, I couldn’t let go of this sunk cost; otherwise, I’d have switched to computer science for my master’s degree.

So I kept studying medicine and hacking computers simultaneously—until nearing the end of my master’s, when I realized I’d accomplished virtually nothing. Gradually, it dawned on me: I couldn’t excel at both TCM and computer science. With my abilities, trying to do both would only lead to burnout—and end with me becoming a mediocre TCM practitioner or a mediocre software engineer. I’d look like a “slash” professional on paper, yet be uncompetitive in either field.

So which path should I choose? I consulted countless people: professors, senior students, family, friends, insiders and outsiders alike—in short, anyone around me whom I considered wise.

The good news? Everyone made reasonable points. The bad news? Their advice was all over the map. Stuck in indecision through the second half of my second year of graduate studies, I finally grasped the truth: this was like “The Little Horse Crossing the River”—only by stepping into the water myself could I gauge its depth.

• The TCM river: I’d already waded in. I’m confident that—even if I never become an excellent TCM practitioner—I can become a competent one, one capable of earning a stable livelihood. (Because most TCM practitioners ordinary people encounter are truly subpar; among that crowd, I’m relatively tall.) My floor is crystal clear: even at my worst, I won’t starve.
• The computer science river: I had no idea whether I’d drown. So—let’s jump in and find out!

Thus, early in 2025, I decided to test myself via innovation-and-entrepreneurship competitions: Could I start from zero and build a product, assemble a team, acquire users, and generate revenue? Could I achieve recognition from myself, official channels, and the market—all starting from scratch?

That led to the “Shitu TCM Input Method” project, which largely achieved all those goals (a fascinating story in itself—I’ll share it later). During this process, I deeply experienced AI’s advantages in programming: AI wrote the vast majority of features; my main tasks were conceptualizing functions and pointing out bugs.

This test revealed several key strengths:

• Team leadership: I can quickly identify talent, assemble teams, and steer them toward the right direction. Many people struggle to recognize talent—for example, assigning a potential art director to software development, or a future sales champion to poster design—and fail to adjust responsibilities promptly based on workflow feedback.
• Idea generation: I can brainstorm numerous feature requirements because I use my own products intensively and improve them based on personal needs. Many creators don’t even use their own products, proposing seemingly logical requirements that turn out meaningless after implementation—wasting everyone’s energy.
• Aesthetic sense: Having consumed vast amounts of “beautiful” content and used countless software applications, I intuitively grasp what’s visually appealing and user-friendly—enabling me to offer meaningful improvement suggestions.
• Communication skills: Years as a livestream host mean I’m never nervous speaking publicly—I can chat comfortably in any setting, and if the topic is within my expertise, I’ll talk nonstop. Many people freeze the moment they open their mouths.
• Rapid learning: Though not a CS major, I can quickly implement new features using video/text tutorials + AI assistance—and roughly understand how they work. When bugs arise, I can usually pinpoint their root cause.

In short: I tested the river—and discovered I won’t drown.

Yet, to some extent, university entrepreneurship competitions are little more than “playing house” among students. How much credibility does this test hold? Unclear. So I moved to the next phase.

I interned for several months at a leading medical AI company—from knowing absolutely nothing to handling backend, frontend, and product work, and eventually interfacing with clients and managing other interns and full-time employees (on this project). This real-world environment confirmed my strengths do translate to tangible productivity gains.

I also joined a genuine hackathon, collaborating with top-tier software engineering talent I could access. I led the entire project’s direction—and once again validated my strengths.

At this point, I was certain: I wouldn’t drown in the real-world river—I’d swim confidently, even joyfully.

By now, I’d clearly defined both my floor and the ceiling above it: I’m confident I can become an upper-mid-tier developer—and have even glimpsed a higher ceiling: I likely have the capacity to lead, managing a small team without issue.

Brimming with confidence, I began job hunting—only for reality to remind me life is full of ups and downs. Rejection came swiftly.

So, Gul’dan—what is the price?

My resume targets were crystal clear: roles bridging medicine + AI + management.

I have a medical background, hands-on experience building software from zero, and extensive communication practice. I understand user needs and technical implementation details. My self-perception? A skilled team “lubricant” and conductor—management roles would maximize my strengths.

As for weaknesses: my formal CS foundation is weak, and I lack relevant degrees or certifications. So I deliberately avoided algorithm or pure development roles.

I applied heavily to product manager positions. All major tech firms rejected me. A common reason emerged: “We need candidates with prior product experience.”

Uh-oh. Because I’d leveled up sequentially—backend → frontend → product—my actual product experience was concentrated in the internship’s final stage, leaving me genuinely underqualified. Catching up now was impossible. Yet I remained convinced my strengths were real—and worth creating opportunities to deploy. So I kept applying, lowering my sights to smaller companies.

What about ByteDance?

After all that preamble—we’ve finally reached ByteDance.

Another application target was healthcare divisions at major tech firms: Alibaba Health, JD Health, etc. One such application landed me at ByteDance’s Xiaohe Health—specifically, its medical model evaluation team. I chatted smoothly with HR and quickly advanced to technical interviews, facing someone directly involved in the business.

First came self-introduction, then detailed questions about experiences listed on my resume—especially large-model projects. Up to this point, I felt I’d performed well. Then we shifted to model evaluation topics—a domain I’d barely touched. My answers were disastrous. One question remains etched in my memory:

If tasked with evaluating a model, how would you proceed? What steps would you take? Which metrics would you use?

I failed the interview. Feedback arrived:

Other candidates in our pipeline have direct, relevant experience, so we’re prioritizing them for subsequent rounds. Your evaluation isn’t negative—just lacking in experience relevance.

Ironically, during my internship, I had been exposed to a model evaluation task—but didn’t prioritize it, delegating it to others and missing a crucial learning opportunity.

Coincidentally, my next ByteDance opportunity centered on another overlooked project: a Douyin medical content moderation team aiming to automate reviews via AI—requiring agent construction and model training. I’m familiar with agent building, but only dabbled in model training. Predictably, I floundered answering related questions during the interview .

Both rejections echoed the same root cause: “We need candidates with experience.”

From the company’s perspective: ideal hires can start working immediately; those with related experience are acceptable; but training newcomers from scratch? Sorry—we’ve got plenty of qualified candidates lining up.

From my perspective: though I possess experience and capability, it doesn’t align with these interviewers’ specific needs. The solution is written on the problem itself: had I deeply engaged with model evaluation or model training during my internship, my success rate in these two interviews would’ve been significantly higher. But I didn’t foresee these roles coming—I’d prioritized other directions, failing to plan my learning path. When encountering unfamiliar topics, I’d try to learn everything, repeating my earlier mistake: With my abilities, I can’t master both A and B simultaneously—and B is precisely the skill most critical for the next 3–5 years. By focusing on A, I let B fall behind.

This lesson taught me: in future full-time roles, I’ll begin by consulting and planning—not diving headfirst and solving problems reactively.

Postscript

Have you ever met people who know exactly what they want?
They race like Kuafu—chasing their own sun,
forward, forward, never stopping.

Once, I couldn’t comprehend how they summoned such energy.
Now, I’m slowly becoming one of them.

On the path toward my goal:
Wind feels exhilarating. Sunlight feels exhilarating. Rain feels exhilarating.
Even if I trip and fall—I’ll laugh, rise, and keep running.