Skip to main content

Engineering team organization

In a 2026 podcast (42章经 × 魏小康), a former hiring leader who worked at both ByteDance (2017–2020, through Douyin’s explosive growth) and Meituan (2020–2024, as hiring lead and AI product manager) shared hard-won lessons on how two of China’s most successful tech companies build and run engineering organizations.

Two organizational philosophies

ByteDance and Meituan represent fundamentally different models: The key insight: neither model is universally better. The right choice depends on what you are building. AI application-layer products that require heavy delivery and integration follow the Meituan pattern: “You have to make it hard, tiring, and heavy.”

Hiring principles

721: battlefield as training ground

Meituan’s philosophy on talent development:
  • 70% — learning by fighting real battles (giving people meaningful ownership)
  • 20% — learning from skilled practitioners (apprenticeship)
  • 10% — formal training
“The most important thing is to give people a battlefield. Good people will fight their way out on their own.”
This is not “we don’t train people.” It is “the battlefield is the training.”

Hire elite, pay premium

ByteDance’s salary strategy: market rate 100, typical job-hop offer 120–130. ByteDance offered 140–150 plus overtime. Pinduoduo went further at 170–180 plus six-day weeks. The logic: “Hiring one top person to solve a business problem costs less than hiring a bunch of people.” From an hourly-rate perspective, the premium is justified.

Expand supply through networks, not headhunters

Startups lack brand influence. The primary hiring channel is warm referrals from trusted people. Treat your best people as a CRM pipeline — the podcast host spent two and a half years meeting a target executive every three months, starting from the day the person joined a competitor, before finally hiring them.
“If there’s a competent person around you, go get all the competent people around them.”
Reverse-network hiring: don’t just recruit one person. Recruit their entire circle.

Today is the cheapest day to hire

“Every day after today will be harder to hire. Whatever you pay today is a bargain. It’s not ‘too expensive now’ — it’s ‘more expensive later.’”
Talent supply-demand tension is structural and worsening. Early investment in hiring compounds.

Communication overhead

“Ten people will get at least 10% of the information wrong.”
Three people working on the same thing already understand it differently. Ten people, and everyone gives a different answer about what the company is doing. ByteDance spent significant time clarifying OKRs — not as bureaucracy, but as loss prevention.

Culture = founder behavior

“Startups don’t need to ‘build culture.’ All top companies have essentially the same culture. The founder’s way of working is the company’s way of working. Just shape a good atmosphere.”
Culture is not a document or a workshop. It is what the founder does every day, observed and replicated.

AI-era organizational shifts

The podcast noted early signals of change:
  • AI application layers demand operational depth over pure model intelligence
  • Delivery and integration work cannot be fully automated away
  • Teams that combine domain expertise with AI tooling will have an edge over pure AI-native startups without industry knowledge

References

Last modified on July 1, 2026