news

The battle for the top 5% of talent

Author | Wei Linhua

Cover source | Movie "Silicon Valley"

During the "golden March and silver April" recruitment season, there are many "help posts" sharing offer information on social media. The posts generally only disclose the company name and salary range, and blur the specific position as a criterion for judging whether it is worth going.

Bai Ding (pseudonym), a 985 master's student graduating in 2024, posted 15 offers for algorithm-related positions he received in the autumn and spring recruitments. "I caught up with the trend of big models, and all companies offered generous offers." Bai Ding wrote in the post that he received a salary that he had never imagined before: the highest offer was an annual salary of 800,000 yuan.

In the comment section, some people questioned whether his experience was fabricated; one respondent was a PhD who felt that his salary was "inverted" in comparison; and some took the opportunity to ask for internship tips.

"2024 is the best year to enter a large model algorithm position," read the title of one post. A few people agreed with it, while the majority criticized it for "inciting anxiety."

In the stories of riding the wind, some people have multiple offers and can get an annual salary of nearly one million right after graduation; some people submit more than 400 resumes in a week, but only receive sporadic interview invitations.

A polarized job-hunting world is showing people its less-than-ideal side.

"The Pig That Takes Off on the Model Wind"

Bai Ding, who just graduated this year, jokingly called himself a "pig that took off in the model boom." Looking back on his campus recruitment experiences, he attributed one of the reasons why he received so many offers to luck.

Bai Ding's original career choice was a traditional algorithm application position, such as search recommendation. While still in graduate school, Bai Ding knew clearly that his research field was relatively niche and it would be difficult to find a job that matched his field.

The popularity of ChatGPT changed his career path.

This year, it seems that all companies are paying attention to AI, and "every company lacks people who can build big models."

Fresh graduates who have just graduated from school are the beginning of the big companies’ scramble for the best. Some big companies have already opened separate recruitment channels for AI technical talents, such as Baidu’s “AIDU Plan” and Meituan’s “Beidou Plan”. This year, Tencent’s “Qingyun Plan” also set up a separate recruitment for AI big models, claiming that the recruitment expansion will exceed 50%; Byte’s Doubao Big Model Team set up the “Top Seed” seed plan, targeting fresh doctoral graduates.

Luo Xing (pseudonym), who works as an algorithm engineer in a large company, also decided to change jobs this year. Compared with the "forced job-hopping" two years ago, this time, he clearly felt that the employment environment has warmed up.

In terms of the number of positions, the number of available positions has increased significantly. Luo Xing told Xuebao Finance that most of these new jobs are related to large models. One of Luo Xing's juniors is a doctoral student who graduated this year. Through the talent plan of a large company, he has received an offer with an annual salary of nearly 1 million yuan.

The AI ​​whirlwind that started with the popularity of ChatGPT at the end of 2022 has spawned a large number of new jobs.

The "2024 First Quarter Employment Big Data Insight Report" released by Liepin Big Data Research Institute shows that in the quarter, the number of generative AI-related positions in China increased by 321.7% year-on-year. Among them, the high-salary segment of more than 500,000 yuan grew the fastest, exceeding 500%.

Judging from the distribution of new positions in the AI ​​field, technical positions dominate. Algorithm engineers top the list with a proportion of 19.30%, followed by product managers and natural language processing.

The narrow and crowded track was widened by new opportunities, and Bai Ding adjusted his career direction accordingly: he focused on large models.

Opportunities kept coming. Before graduation, Bai Ding received 15 offers from many large companies including Baidu, ByteDance, and Alibaba, with annual salaries ranging from 500,000 to 800,000 yuan. Bai Ding had plenty of room to choose from, and he even received offers from multiple departments in the same large company.

After two weeks of continuous interviews, Luo Xing received six offers from large companies, with annual salary packages ranging from 700,000 to 800,000 yuan. In the end, he chose to go south to join a large company in Hangzhou.

"We can afford anything that others can afford"

Behind the million-dollar annual salary, large companies have a set of strict screening standards: in special recruitment plans, master's and doctoral degrees are the basic threshold for most companies, and "super academic masters" and "competition masters" are plus points.

"It seems that large companies only recruit the top 5% of graduate students for algorithm research," Siman (pseudonym), a graduate student at the University of Manchester, told Snow Leopard Finance. "The requirements for algorithms in large companies are really too high."

In fact, in the promotional video of ByteDance’s “Top Seed” recruitment plan, a business leader has made it clear: “We need the top 5% of people to accomplish things that the 95% of people cannot accomplish.”

"The people we are looking at must be the top 5% of people" Image source: "Doubao Big Model Team" video account

Siman, who will graduate in September this year, is preparing for the autumn recruitment. She complained on social media that it is too difficult to get into a large company. In the comment section, she received two completely different voices. One voice told her that this autumn recruitment is the easiest year to enter the algorithm industry; the other voice said that if you don’t have outstanding conditions, it is basically impossible to enter a large company this year.

The latter also added that the recruitment standards of large companies are at least "double 985 degrees and papers published in top conferences and journals." This made Sman sigh that those who can publish papers in top conferences are basically "top students who will be sought after even if they are studying for a doctorate."

Bai Ding recalled that there were many better candidates in the same interview batch as him, and some of them might have published seven or eight top conference papers, "but this is not enough to ensure that they will pass the screening."

A headhunter told Snow Leopard Finance that for campus recruitment positions for large model algorithm engineers, large companies offer monthly salaries in the range of 30,000 to 60,000 yuan; for social recruitment positions, they can basically offer an increase of about 40% to 50%.

Based on academic qualifications, internship/work experience, scientific research results and project experience, the offers made by companies are clearly stratified.

The highest level of offers comes from top talent programs set up by various large companies. Taking Huawei's "Genius Youth" program as an example, the lowest level is about 800,000, the next level can double to about 1.5 million, and the highest level reaches 2 million per year.

In terms of campus recruitment, even the same position is divided into three levels: SSP (Super Special Offer), SP (Special Offer) and "bargain offer".

There is a huge gap in salary between different levels. Bai Ding received multiple offers for SSP positions, with an annual salary of about 700,000. One of his seniors chose to join the same large company, with an annual salary of only 400,000, which is nearly half the difference. "Take Alibaba as an example, the cheapest salary for algorithm positions is around 400,000, which means you only get 26k a month," Bai Ding added.

Although most of these offers are negotiated on a face-to-face basis, and it is impossible to get an accurate range, "high salary" is a standard feature of the promotion. Meituan's "Beidou Project" loudly declared on its campus recruitment poster: "What other companies can afford, we can afford."

Many headhunters told Snow Leopard Finance that the top talents "can get at least seven or eight offers a month." Luo Xing's experience is that "when HR knows that you have offers from other big companies, they will generally re-determine your salary."

472 deliveries that failed to make a splash

Top talents hold multiple tickets to the pyramid at the same time. However, the people who are being fought over are always a minority.

Siman, who is about to graduate with a master's degree, has already felt the chill of this year's autumn recruitment season. In just one week, she submitted a total of 472 resumes, but received only 30 to 40 replies that week, and only received interview invitations in the single digits.

"At first I thought there was something wrong with my resume, but in the end I realized it was the person who had the problem," said Siman, who did not dare to apply for talent programs at large companies, as she felt her chances were slim.

From last year's autumn recruitment to this year's spring recruitment, Che Wen (pseudonym), a graduate student who will graduate this year, applied to all the Internet and car companies he could. "At first I thought I could at least get into a medium-sized company, but later I found it was almost impossible," said Che Wen.

The resumes he sent out either failed the initial screening or were rejected in at least three rounds of interviews. After a year, he only received offers from Ideal and Huawei, but he gave up both because the salary did not meet his expectations. "21k, even undergraduates can get the same treatment." Che Wen was a little disappointed.

A small number of job seekers have taken off on the AI ​​wave, but on the other hand, even fresh doctoral graduates from well-known universities cannot get the offers they want. Liepin data shows that the number of generative AI positions has increased by more than 300%, but the number of job seekers has increased by more than 900%. With limited positions, job seekers still face the problem of too many people and too little porridge.

Many job seekers told Snow Leopard Finance that ordinary-sized companies lack computing resources and startups lack stability, so most people's attention is focused on large companies.

Talent gaps still exist in large companies, but these vacant positions would rather be retained than released easily. After Luo Xing joined the company, his department continued to recruit. "Although the desire to recruit people is quite urgent, it is not easy to find the right person." In the past three months, only one candidate joined his department.

Bai Ding feels that compared with last year, the interview threshold for large model-related positions has become higher. Without relevant experience, it is even difficult to find a suitable internship in a large company.

Several headhunters told Snow Leopard Finance that during the campus recruitment stage, academic qualifications, scientific research achievements and are key indicators for measuring talent. A master's degree is the basic threshold, while top positions only require a doctoral degree. For social recruitment, companies also need to examine the match between personal project experience and recruitment needs.

As graduation day approaches, Siman has revised her resume several times. She has refined her project experience and is looking at more positions in R&D, testing, and robotics. Siman is no longer as picky about companies as before. While learning new content, she applies for all the positions she can find.

Che Wen sticks to the goal of last year, still looking for opportunities in algorithms, autonomous driving and other fields, and is preparing to join the new round of autumn recruitment. "I can't change the environment, all I can do is keep submitting resumes," said Che Wen.