.png)
4 hours ago
1
Note: The below text is a translation of a submission to ichaoqi.com
Yesterday, I received an unusual private message.
The sender claimed to be a former ByteDance employee. He wanted to submit a series of articles written by his friend, documenting their firsthand experience working in ByteDance’s U.S. e-commerce unit. He hoped I could help share them with a broader audience.
Truthfully, I was conflicted.
I cannot guarantee that what’s written here is “the truth.”
As someone not directly involved, I lack the means to verify whether the highly specific and detailed accounts are 100 percent accurate. From my years of reporting on the tech industry, I know that single-perspective narratives inevitably contain bias. They represent a position, not necessarily the truth.
I, Ni Shu, offer no endorsement of the content that follows.
I am sharing it for one reason alone: that one line—
“Please tell me the world isn’t entirely dark.”
I believe freedom of expression exists in modern Chinese society.
I believe that hearing multiple sides of a story brings us closer to the truth.
I ask readers to read with discernment:
Which parts are real, and which are not?
Published: March 31, 2025
In mid-2022, I made a bold move: after nearly eight years at Amazon, I joined ByteDance’s international e-commerce team in the United States. I was full of hope, believing I was stepping out of my comfort zone to become a pioneer on a new frontier. Little did I know, it was the start of a long journey into disillusionment.
I had spent nearly eight years at Amazon, over six of them based in the U.S., and had long secured my residency status. I was a Level 6 Manager with a team of more than a dozen people. My direct and skip-level managers treated me well, and my performance ratings had been solid—“Exceeds” two years in a row in 2020 and 2021 (not to brag, just to highlight how ironic my later performance review at ByteDance would be). My manager had even committed to helping me push for a promotion to Level 7.
But reality was sobering: the business unit I was in was stable and low-growth, unlike Alexa or other aggressively expanding divisions. With my manager and skip already at L7, there was little room for upward movement. I couldn’t stay at Amazon forever, right?
By then, the winds in tech were already shifting. Layoffs and hiring freezes were becoming more common. I still got interviews when I sent out résumés, but most companies said, “No headcount” by the time the interviews rolled around.
I interviewed with three companies. At Uber, I had a great technical phone screen with an Indian director, but got rejected the next day without explanation. At Google, things were worse—five scheduled virtual interviews, three of which were no-shows. When I finally did the makeup rounds, my mindset was already wrecked, and I bombed both the coding and system design interviews. To top it off, they threw in a machine learning round—completely unexpected for a Software Development Manager role. The HR hadn’t even mentioned it. That whole experience could be its own article: “Interview Trauma Chronicles.”
Just as I was resigning myself to staying put, I saw a post on LinkedIn: ByteDance was hiring an e-commerce manager.
With eight years in Amazon’s e-commerce trenches, this was clearly my arena.
After submitting my application, the HR response came at lightning speed—like someone had hit 2x playback. They scheduled a phone screen that day. It was just a casual conversation about my background and motivation. A few days later, I was told to prepare for a virtual interview: three to four rounds.
The interviewers—let’s call them H, Y, and XF—were all referred to as “classmates,” a ByteDance quirk. First up was H, then Y, then XF, and a fourth round I barely remember—likely another HR rep.
Only later did I learn that XF was the head of international e-commerce engineering. H and Y reported directly to him and held fairly senior roles (ByteDance’s title system is opaque, but XF was likely at level 4-2, and H and Y at 4-1).
The interview focused on my experience at Amazon—technical and business details, yes, but none of the Big Tech “eight-legged essays.” No coding, no system design. A few days later, HR said they were ready to make an offer. I was surprised—was hiring a manager in a Chinese company really this easy?
I later realized I was a special case: part of the first wave of hires, with the team in urgent need of talent, and my e-commerce background and bilingual skills made me a perfect fit.
These three interviewers would later become central figures in my story. To introduce them briefly:
XF was the head of engineering for international e-commerce.
H oversaw all U.S. e-commerce engineering and was intended to become a virtual dotted-line manager for all local leads.
Y led a relatively independent sub-organization. After some internal discussions, they placed me in Y’s department, making him my direct manager.
Before signing, I had a one-on-one with Y to clarify the role and growth opportunities. He said I would lead the U.S. engineering team under his department, at level 3-2 (equivalent to a Senior Manager), with plans to build a team of about 40 people. The scope sounded solid.
Compensation, of course, was a key concern—I was supporting a family of four. ByteDance had no RSU buyback plan in the U.S. at the time, and an IPO was nowhere in sight. I asked to increase the cash portion, essentially treating the equity as worthless for now. If the stock eventually became liquid, great—it’d be a lucky windfall.
After brief negotiations, HR bumped up the base salary and signing bonus as requested, but wouldn’t increase the stock. Fair enough. My RSU grant was valued at $195 per share, but post-hiring, the valuation dropped to around $150–160 due to the broader market downturn. I joked that I had “bought stock at the peak with my own body.”
Still, the total comp beat Amazon’s. Add to that a sense of purpose—wanting to contribute to a Chinese company—and the appeal of working in my native language, plus a weak job market, and I signed.
In October 2022, I officially joined ByteDance.
I was full of ambition, convinced I wasn’t just changing jobs—I was stepping onto a bigger stage.
Who would’ve thought this was just the opening scene of a long and dramatic saga?
Published: April 2, 2025
In October 2022, I joined ByteDance’s international e-commerce team in the U.S. with boundless energy. COVID had just subsided. Our Seattle-area office in Bellevue hadn’t opened yet, and my work laptop was shipped directly to my home. On my first day, I assumed I might be the first employee onboard—only to find that three “classmates” had already beaten me to it.
Those early days were a whirlwind of optimism and energy, like the start of a scrappy startup adventure.
When I joined, there were already three others: LB, WT, and SG. They had already begun a partial return-to-office rhythm (RTO2)—two days a week in person—at a WeWork space in Bellevue. I met them in my second week, on a Tuesday.
After nearly two years of remote-only work, I was thrilled to finally see colleagues face-to-face. We weren’t just a thrown-together squad—we were a small but mighty team, each of us above level 2-2, all seasoned professionals eager to make our mark.
The team’s trajectories diverged in fascinating ways. SG led a promising business direction and built a team of over ten, but a later reorg reassigned him and split us apart. LB, widely acknowledged as the hardest-working, soon took technical ownership of a major project and became a pillar of the team. WT, though initially promising, struggled to keep pace and eventually left. Our four-person "founders' team" had all the drama of a workplace drama: some soared, others faded quietly into the background.
One of my top priorities was hiring and building the team. Y, my manager, gave me a “small goal”: build a 40-person team within six months. That meant hiring over six people a month. Based on HR data from other U.S. teams, this would require screening more than 600 candidates.
At the end of 2022, while the rest of Big Tech was laying off and freezing hiring, ByteDance bucked the trend and became a rare hot spot for job seekers. I was interviewing candidates almost daily and running paid job ads on LinkedIn. In just a few days, my LinkedIn connections jumped from a few hundred to over a thousand. My inbox exploded with messages I couldn’t keep up with.
I personally referred 300–400 candidates—but only 5 or 6 made it through, since I was only pushing them to my own team and earned no referral bonuses. It felt like being a “public servant.” Still, I took pride in building a team from scratch. For more senior roles, we relied on company sourcers and headhunters. Most junior candidates I found via LinkedIn—mostly recent grads or those with limited experience.
In just three months, our team grew to more than ten people—making us the fastest-growing engineering team in ByteDance’s U.S. e-commerce unit. We hit multiple milestones in rapid succession: the first frontend engineer, the first woman, the first non-Chinese team member, the first campus hire. I became something of an “interviewing machine,” breaking the 100-interview mark with ease.
As for the actual business—more on that later. Our focus was building features tailored for the U.S. market.
ByteDance’s hiring process differed noticeably from American tech firms. For engineers in the U.S., companies usually begin with a technical phone screen focused on coding. ByteDance largely skipped this (except for campus hires, who had online tests). After an initial HR chat, candidates went straight to virtual interviews (VOs), scheduled one by one rather than bundled in a single day. Only after passing each round would they move on to the next. The final round was the “leveling” interview, where a manager or senior lead would make the final call.
Initially, I didn’t have the authority to set levels. So after I interviewed candidates, I had to escalate them to Y for review. Later, we realized this was too slow and created a poor experience. Eventually, I was granted 1-2 and 2-1 leveling authority, allowing me to finalize levels for non-senior hires in three rounds. After passing the 100-interview mark, I was granted 2-2 authority, which sped things up even more.
At ByteDance, managers must complete a formal “management presentation” to receive leveling authority. This is a unique feature of ByteDance—typically done six to twelve months after joining. It involves summarizing business results and team plans in a formal pitch to one’s manager and partner leads, followed by live Q&A. It’s not a formality—many fail. Since our U.S. team was new and lacked deep business traction, no one pushed hard for the review deadline. I began preparing, but then a massive reorg hit, my scope changed, my reporting line shifted, and the presentation never happened.
Without completing that review, a manager technically can’t manage people directly. But in the U.S., where localization and diversity were top priorities, this rule couldn’t be rigid. HR even tracked the ratio of non-Chinese employees per team. Having everyone report back to a domestic lead would have been a logistical nightmare. So we local team leads were given special permission to manage directly. ByteDance’s rules were rigid on paper, but flexible in execution—everything depended on timing and internal politics.
Those early months felt like drinking adrenaline: the pace, the urgency, the pride of creation. Our four-person squad at WeWork coded, documented, and interviewed nonstop. We grew from zero to 10+ people amid laughter, exhaustion, and raw excitement.
I believed I’d finally found my grand stage.
How far could we go?
I didn’t stop to ask. I just ran forward.
Published: April 3, 2025
While growing the team, we also plunged into building ByteDance’s U.S. e-commerce platform from the ground up. But there were two massive hurdles we could never avoid: the Trusted Third Party (TTP) environment and the U.S. Data Security (USDS) team. These weren’t just compliance frameworks—they were operational landmines.
TikTok’s e-commerce business in the U.S. was subject to intense scrutiny, especially regarding data security. ByteDance invested hundreds of millions to build a fully isolated environment with Oracle, dubbed TTP (Trusted Third Party), to reassure U.S. regulators.
In theory, it was a privacy safeguard. In practice, it was a developer’s nightmare.
Every piece of data leaving the TTP environment had to be flagged—what could go out, what couldn’t. Each data tag had to be reviewed and approved by the security team. To view logs or debug production issues? You had to submit a ticket to USDS, which would manually screen sensitive info before sharing anything back.
A single bug could take days to troubleshoot.
USDS wouldn’t work overtime unless it was a P0 production issue—the highest possible priority. Tight deadlines? Doesn’t matter. You had to negotiate staffing with their team lead in advance.
We estimated that launching the same feature in the U.S. took 50% more effort and nearly double the time compared to China. TTP turned development into a test of stamina.
And then there was Oracle Cloud.
Its availability was nothing like my former employer, AWS. In summer 2023, a heat wave knocked out several services. The root cause? Not enough fans in the server rooms. They didn’t have backups either. It wasn’t until the evening—when temperatures dropped—that things slowly came back online.
Another time, latency across our services spiked dramatically. The reason? Someone manually unplugged a network cable while switching routers, causing a traffic spike and packet loss. It took hours to recover. These incidents were as infuriating as they were absurd.
USDS was a mysterious organization. Its members had to be “clean”—initially only U.S. citizens or green card holders from “friendly” nations were allowed. Even later, it remained tightly controlled.
Each USDS member supported multiple teams and had zero business context. You had to spoon-feed them.
Need database access? You had to write out the exact SQL in a ticket. They’d copy-paste it, run it after filtering sensitive fields, and send you the output. One typo in your query could cost you days.
One incident captured the dysfunction perfectly.
We’d just onboarded a third-party vendor for merchant verification when, suddenly, they pulled out. The reason? Someone from USDS had contacted them directly, warning them that signing with ByteDance was risky and suggesting they sign with USDS instead. They framed it as a customer data issue.
But merchant data wasn’t user data. And USDS was still ByteDance.
The vendor panicked and backed off. The whistleblower? Faced zero consequences. I was stunned.
Is this how privilege plays out when Americans run compliance on American soil?
Despite these compliance handcuffs, we had to meet hard deadlines for the U.S. Go-To-Market (GTM) plan. The launch was staged: first Beta, then Gamma, and finally full GTM.
We had to back-plan every required feature—first for Beta, then Gamma, then launch—while working with domestic e-commerce teams who were all based in China. Cross-time-zone meetings became a daily struggle. Working until 2 or 3 a.m. was routine. Everyone was burned out, like overworked cotton ginners.
Despite the grind, our efforts paid off.
By 2023, we had rolled out a series of new features tailored to U.S. users, gradually moving from closed beta to semi-public to full GTM. Watching GMV (Gross Merchandise Volume) climb from zero was like seeing our own crops blossom. We were exhausted—but proud.
Published: April 5, 2025
Soon after joining ByteDance, I realized its approach to goal-setting and project management came with its own unique “flavor”—frantic, relentless, and often baffling. OKRs and priority labels like “P0” weren’t just directional tools—they were handcuffs. What follows is a glimpse into the chaotic sprint of “embracing change” at ByteDance.
When I first joined, OKRs were set every two months. Yes, you read that right—bi-monthly.
Just as you finished one cycle, you barely caught your breath before the next began. Everyone was miserable. After months of pushback, HQ finally switched to quarterly OKRs.
Even quarterly cycles are considered fast by most standards—at Amazon, once-a-year OP1/OP2 planning was sufficient. At ByteDance, the OKR system mainly targeted managers and key contributors. For rank-and-file employees, the pressure was lighter.
Why? Because in a place where projects change daily, how can frontline staff write meaningful OKRs?
You might set an OKR to finish Project A, only to have it replaced by Project B next week. Writing OKRs often felt like a pointless ritual.
After OKRs came the quarterly project prioritization.
Resources were always scarce. But in a new business like e-commerce, the demands were infinite. So every business unit fought to label its projects with the highest urgency.
P0 was supposed to mean “top priority.” But then someone got creative—P00 entered the scene, shorthand for “top of the top.” And soon after, P000 showed up. Leadership finally stepped in, banning anything beyond P0. Left unchecked, we joked, P0000 would’ve been next.
Still, not long after, P00 quietly came back. Thankfully, P000 didn’t.
You had to listen carefully in meetings—one zero too many, and you’d misunderstand the stakes entirely.
The title “P00” didn’t automatically grant you power. P00s typically came from leadership’s OKRs.
Each org submitted its high-priority projects for consideration. Then, leaders across business, product, and engineering would “select” which ones deserved P00 status and the resources that came with it. It felt like a royal selection ceremony: the projects were lined up like concubines, and the emperors picked their favorites.
But even among P00s, there was a hierarchy.
In a business-driven environment like e-commerce, the business lead’s P00 was king. Our business lead—let’s call him F, affectionately known as “Teacher F”—wielded his P00s like imperial edicts. Once his seal was on a project, all teams—product, engineering, and ops—dropped everything to execute, even if it meant all-nighters and reverse-planning with zero margin.
I remember one of Teacher F’s P00s estimated by a China-based team to take 21 days. Teacher F was displeased and escalated to our engineering VP. That VP promised: “You get seven days.” This included a Mid-Autumn Festival holiday. I’m sure that team turned pale on the spot.
And even once a P00 was locked in, emergency “insertions” from Teacher F could reshuffle priorities again—someone’s project would get cut without warning.
We got used to asking, “Whose P00 is this?” Because if it was Teacher F’s, we knew it meant sleepless nights.
OKRs and project prioritization kept us running nonstop.
Switching from bi-monthly to quarterly felt like catching a breath, but P00/P000 chaos, last-minute demand shifts, and Teacher F’s imperial projects pulled us back into “ByteDance speed.”
In the moment, it felt glorious—like we were doing something that mattered.
But looking back, many projects were written on Monday, changed on Wednesday, and abandoned by Friday. Some were completed but never launched.
The fatigue crept in.
This chaos—framed as “embracing change”—revealed deeper cultural rifts between Chinese and American management styles.
Project flow, decision-making, and even leadership philosophy were worlds apart from what I’d experienced in the U.S.
Published: April 7, 2025
At first, I was full of admiration for ByteDance’s speed. But over time, I began to see the other side of that coin: short-term thinking, fragmented execution, and a top-down culture that often felt out of sync with how American teams operate.
This tension—the slow burn of a deep cultural divide—became one of the most defining challenges of ByteDance’s U.S. e-commerce venture.
There’s a saying inside ByteDance: ByteDance speed is faster than light speed. That might sound like a joke, but the execution expectations were very real.
In China, people regularly pull 996 schedules (9 a.m. to 9 p.m., six days a week), respond to messages at all hours, and expect timelines to shift overnight. The assumption is that engineers should be able to produce MVPs in a week, iterate in two, and scale by month’s end.
In the U.S., things don’t work like that.
People value planning, work-life balance, and sustainable schedules. But the pace set by Beijing didn’t slow down just because we were in a different country.
Our local teams constantly had to “translate” product requirements written in Mandarin, join late-night syncs with Beijing, and deal with changing specs dropped into Slack after midnight.
The result? Burnout. Frustration. Confusion.
We often joked that our job was just to “take the chaos and add subtitles.”
Let’s be honest: Beijing called the shots.
The “localization” plan made nice promises on paper, but strategic direction, funding, and major product decisions all flowed from China. The U.S. teams were implementers, not decision-makers.
Even on things like promotions or performance reviews, the final vote often came from managers in China who barely interacted with us. You’d spend a year leading a complex U.S. launch, but if your Beijing counterpart didn’t “get” your work—or worse, disagreed with your communication style—your review could tank.
I once fought hard for a team member’s promotion. All signals pointed green: impact, peer feedback, scope. But Beijing vetoed it with a vague line: “Not enough leadership visibility.” There was no recourse.
This wasn’t a one-off. It was structural.
Many leaders in Beijing came from strong engineering backgrounds. They were sharp, aggressive, and expected to be the smartest person in every room.
That led to a culture of “manager-as-owner”: your job was to execute, not debate. Managers often wrote specs, set milestones, and even reviewed your code.
In the U.S., especially in tech, the expectation is different. Managers are enablers. They guide but don’t micromanage. They coach but don’t command.
This gap led to tension. American team members felt disrespected or ignored. Chinese leaders saw U.S. teams as “too soft” or “not hungry enough.”
Both sides had a point. But neither adapted fast enough.
Language wasn’t just a logistical issue—it shaped power dynamics.
English-speaking employees often struggled to follow Mandarin-only documents or meetings. Translation tools helped, but nuance got lost.
And Chinese-speaking leaders often defaulted to Mandarin in high-pressure moments, excluding others by accident—or sometimes not.
One of my colleagues, a senior engineer from a non-Chinese background, told me privately: “Sometimes I feel like I’m not even in the meeting.”
That sense of exclusion built up, even among high performers.
ByteDance talked a lot about “inclusivity.” But when things got tense, people reverted to comfort zones—and that often meant Mandarin.
Many U.S. teams believed we needed to slow down and localize properly: understand American consumer behavior, build the right features, follow proper compliance.
Beijing’s view was: ship fast, iterate later. Don’t overthink—outwork the competition.
We built things that looked good on dashboards but didn’t resonate with local users. We pushed growth hacks that got short-term GMV but didn’t build trust.
Sometimes it worked. More often, it didn’t.
But the Chinese teams pointed to the short-term GMV gains and said, “See? It works.”
We didn’t speak the same strategic language.
Published: April 9, 2025
The high I felt when I joined ByteDance didn’t collapse all at once. It eroded—bit by bit—with each reorg, each political maneuver, each unmet promise. By the end, I was no longer building something bold and new. I was just surviving.
ByteDance’s e-commerce unit restructured constantly.
Every few months, a new org chart would appear. Team names changed. Reporting lines shifted. Projects were merged or canceled. Sometimes people were shuffled without even being informed—Slack handles would suddenly show new managers, and you’d have to piece together what happened through backchannels.
At first, I thought: This is a growing company—it’s natural to adapt.
But after the fourth or fifth reorg, I realized it wasn’t growth. It was instability.
Each new leader came with a new vision. Projects were paused mid-flight. Headcounts were frozen, then unfrozen, then slashed. Roadmaps were rewritten before the ink dried.
And in each shuffle, some people gained power, while others disappeared quietly into backwater teams.
Politics slowly became the dominant currency.
Leaders began forming alliances, subtly undercutting others, pushing pet projects forward not because they worked—but because they had internal backers.
A senior colleague once warned me: “If you’re not aligning with someone above you, you’ll get eaten alive.”
It wasn’t paranoia.
I saw sharp, competent team leads pushed out because they weren’t “connected.” I saw underperformers protected because they had someone in Beijing vouching for them.
One leader we’ll call L played this game masterfully. He aligned with a VP in China, got placed in charge of a high-profile initiative, then moved aggressively to pull resources from other teams—including mine.
His project failed, but he landed safely in another role. The teams he cannibalized? Scattered.
We all learned the same lesson: results mattered less than relationships.
I had joined with the promise of building a 40-person team. For a while, it felt like I was getting there.
But after multiple reorgs, my scope shrank. Projects were reassigned. Headcount was cut. Promises to promote my team members were delayed, then forgotten. I fought for them, but I had no real leverage.
Some left quietly. Others stayed but disengaged.
The culture of belief—so strong in the early months—was gone. People stopped talking about building the future. They talked about burnout, transfer opportunities, or how to negotiate better severance.
We started joking that ByteDance’s motto should be “Run at full speed… until we crash.”
I had led one of the earliest, fastest-growing U.S. teams. We delivered results, onboarded major systems, and supported GTM launches. I thought I had earned some credit.
But in my performance review, I was rated “Meets Expectations.”
No explanation. No negative feedback. Just… flat.
Others in similar roles got “Exceeds.” Some got promoted.
I asked why.
The response: “Your impact wasn’t fully recognized across orgs.”
Translation: I didn’t play politics well enough.
That was the moment something broke inside me.
I joined ByteDance because I wanted to build something big, meaningful, even historic. A Chinese company going global—what could be more exciting?
But I watched that dream corrode. Not because we weren’t smart enough. Not because we didn’t work hard enough.
Because we lost our way.
We let speed override strategy. We let politics override talent. We let slogans replace real leadership.
And in the process, we burned through good people like firewood.
Published: April 11, 2025
I left quietly. No LinkedIn post. No farewell emails. No “thank you for the memories” Slack messages.
I walked out the same way I came in—quietly, with hope. Except this time, that hope was gone.
After the disappointing performance review, I knew my time was running out.
My team had been dismantled. My projects were reassigned. The roadmap was increasingly irrelevant to anything I cared about.
I tried to stay professional—support the transition, help teammates transfer, write docs, wrap things up.
But inside, I was already gone.
I was tired. Not just physically, but emotionally—tired of navigating politics, defending my team, explaining U.S. culture, and managing up across time zones.
It wasn’t one bad moment. It was death by a thousand cuts.
When I finally told my manager I was resigning, he wasn’t surprised. “I understand,” he said.
There was no counteroffer. No push to stay. Just HR logistics and a final exit interview with someone who didn’t really know what I had done over the past two and a half years.
I handed in my laptop. Cleared my calendar. Logged out of Lark for the last time.
I felt nothing but relief.
Not because I hated ByteDance.
But because I knew I had nothing left to give.
I don’t regret joining.
In many ways, ByteDance was the most intense and eye-opening experience of my career.
I saw what it meant to scale fast. To ship globally. To fight regulatory battles. To build in public under the scrutiny of governments, competitors, and skeptical users.
I worked with brilliant engineers, fearless product managers, and scrappy operators who moved mountains on pure willpower.
But I also saw what happens when ambition outpaces structure. When short-term wins mask long-term rot. When culture becomes fragmented and leadership loses the plot.
I learned that speed is a weapon—but also a poison.
That politics, once tolerated, spreads like rust.
That even the smartest people in the world can fail if they forget what they’re building, and why.
ByteDance’s U.S. e-commerce business is still running. Maybe it’ll succeed. Maybe it won’t.
But whatever happens, I hope it remembers something:
Startups don’t die because they move too slowly.
They die when they move too fast in the wrong direction—and refuse to stop.
Postscript from Ni Shu:
After this story was published, many readers messaged me with a single reaction: “Felt like I was reading about my own company.”
That’s the power of stories—not because they’re objective truths, but because they capture something deeper: the shared experience of ambition, burnout, disillusionment, and quiet resilience.
To the author: thank you for telling your truth.
To the reader: may we all find workplaces that honor both our effort and our humanity.
