跪拜 Guibai
← All articles
Full Stack · Agent · AI Programming

Six Months Unemployed, Then a 15% Pay Cut: A Frontend Engineer's Job Search Postmortem

By 双越AI_club ·
Read original on juejin.cn ↗ Google Translate ↗ Alt translation

The six-month gap ended with an offer, but at a 15% lower salary and only after the engineer stopped restricting his search to Shanghai. Algorithm drills matter less at smaller companies, while Agent and full-stack skills are becoming the new baseline for frontend roles in China's hiring market.

Summary

A Shanghai frontend engineer with five-plus years of experience was laid off in April and spent six months getting back into the market. He spent the first month preparing for big-tech interviews and failed every one, because algorithm and principle questions caught him unprepared. The next two months went to reviewing algorithms, internals, and framework source code, but June and July are the interview off-season and almost no invitations came. When applications resumed in late August, he got 7-8 interviews and converted two into offers, accepting one in Hangzhou at a 15% salary cut.

His own postmortem: monthly application attempts kept feedback flowing, and applying nationwide mattered because Shanghai had few openings while Hangzhou and Shenzhen had more. He regrets spending so much time on algorithm and principle prep, since small and mid-size companies rarely ask those. His advice is to start practicing interview questions and sending resumes earlier, and to take a same-salary offer instead of holding out for a big company or a raise.

He also started learning Agent tooling during the later part of the gap and plans to keep going, concluding that pure frontend roles are gone and full-stack plus Agent skills are the way forward. His plan to master every internal technology, from event tracking to CI/CD to monorepo, and open-source personal projects drew pushback from Shuangyue: promotions depend more on team headcount quotas and KPI scores than on technical breadth.

Takeaways
— A frontend engineer in Shanghai with 5+ years of experience was laid off in April and remained unemployed for six months before accepting an offer in Hangzhou at a 15% lower salary.
— He failed every May interview because he hadn't prepared for algorithm questions and principle questions.
— He spent two months reviewing algorithms, principles, and framework source code, but June and July produced almost no interview invitations.
— After resuming applications in late August, he received 7-8 interview invitations and eventually two offers.
— He applied nationwide because Shanghai had few openings; Hangzhou and Shenzhen had more.
— He concluded that small and mid-size companies rarely ask algorithm or principle questions, so he should have started interview prep and applications earlier instead of aiming for a big company and a raise.
— He believes pure frontend roles no longer exist and that full-stack plus Agent skills are necessary.
— Shuangyue's counterpoint: promotions depend mainly on company/team headcount quotas and KPI scores, not just technical ability.
Conclusions

The 15% pay cut and the shift from Shanghai to Hangzhou show how much leverage employers have gained since the 2020-2021 hiring boom; re-entering after a gap costs salary and location.

The engineer's regret about over-preparing algorithms contradicts the standard big-tech interview playbook, but matches a market where small and mid-size companies dominate openings and care more about practical delivery.

Shuangyue's pushback separates technical skill from career advancement: in a company with fixed promotion quotas, being a broad technologist doesn't guarantee a raise, and treating open-source side projects as a promotion path can be a misdirected effort.

Concepts & terms
Agent
In AI programming, an agent is a system that plans and executes multi-step tasks with limited human supervision. For developers, learning Agent skills usually means building or using AI agents that can read code, run tools, and iterate toward a goal.
埋点 (event tracking)
Instrumentation embedded in frontend or backend code to record user actions, page views, and product metrics. Chinese product teams call this 'burying points' because the tracking code is planted at specific spots.
monorepo
A software development strategy that keeps multiple projects or packages in a single repository, with shared tooling, dependencies, and CI. It is common in large frontend and full-stack codebases.
Source: juejin.cn ↗ Google Translate ↗ Backup ↗