Datawhale Share
Author: Li Bojie, Ph.D., University of Chinese Academy of Sciences
Hello everyone, more and more people are choosing to study for a doctorate. Some people say that a doctorate will not worry about employment, while others say that a doctorate is a waste of time.
Work papers, age anxiety, and most PhD students are no longer young after graduation. What is it like to get a job after graduating with a PhD?
Today I saw a very good doctoral graduation experience sharing on a website. I recommend it to everyone. I hope it will be helpful to friends who have the same confusion.
The "I" below refers to the author Li Bojie himself.
Editor:Jack Cui
Link: https://www.zhihu.com/question/559157484/answer/3149816238
I am a Ph.D. student at the University of Science and Technology of China. When I graduated, I basically got top offers from major manufacturers, including Huawei Genius Boy, Alibaba Star, etc. In the end, I chose Huawei. The following is a sharing of my job search experience.
Graduating with a PhD, academia or industry, starting a business or a large factory?
When I graduated with a Ph.D., it came to job choices, which I roughly divided into five categories: academia, freelance work, entrepreneurship, large foreign companies, and large domestic companies.
The first is academia. Although I like scientific research very much, just like my concerns when I first started studying for a Ph.D., I don’t like theory and like practical research. I feel that research in academia, even in the field of network systems, is often divorced from actual needs. For example, during my Ph.D., the entire team spent several years researching. Although they published good papers, they only explored some possibilities and were not commercialized in any production system. Today, I still hope to summarize the successful practices in the industry and publish them in papers, just like I like to write blogs to share my life, but I don’t want to "force myself to express my sorrow for coming up with new words" and do some short-term work just to write papers. Something that cannot be commercially used.
Second is freelancing. Although the hacker spirit essentially encourages freelancing, the current knowledge-based benefit sharing mechanism is not yet complete. Except for a few big Vs, they only rely on consulting services, knowledge payment, and advertising revenue from personal websites to support themselves. Very difficult. For most freelancers, the money they can earn from freelancing is far less than working for a large factory. As a result, my idea of freelancing slowly faded away. I envy LUG’s friends who choose to work as freelancers and practice the spirit of idealism with their own lives. For now, you still need to ensure that you don’t starve to death before talking about the hacker spirit of freedom, sharing, and openness.
Then there is entrepreneurship. The advantage of startups is that they are highly efficient and can penetrate market segments like catfish. Analyzing the size of the target market and the current player situation, you will find that it is not easy for the company to make money. For a company the size of Microsoft, a market worth less than $1 billion is simply out of reach. Because if a large company wants to establish a project to do something, it requires complex processes and the support of a multi-functional team. What a small team of three or five people can handle, a large company may need a team of dozens of people. The development process of large companies is complex, with a code productivity of 300~500 lines/person-month, and the code produced has been rigorously inspected and tested; a small company may have a code productivity as high as 3000~5000 lines/person-month, but the code quality may not be that high. Therefore, large companies are more suitable for B-to-B markets with strict quality requirements, while startups are more suitable for markets with small steps and fast iterations.
It is more difficult to start a business in the data center network research field where I work, because data centers are only owned by large companies. Large companies either develop their own smart network cards or purchase smart network cards from other large companies. Who would buy an unknown one? What about network cards for small companies? Therefore, although there are many start-up companies in this typical to B field, they are all founded by successful bosses of large companies and are not suitable for young people to start their own businesses. If you want to start a business, you may have to work in relatively "emerging" fields such as blockchain, metaverse, privacy computing, and artificial intelligence. However, I have no experience in these fields, and I don't want to give up my experience in the data center field. .
I think the data center field is still more suitable for working in a large factory. Compared with freelancing or startups, large companies have more rules and regulations. These rules and regulations are reflected in rules and regulations, project management and development processes on the one hand, and in topic selection and project establishment on the other. In the words of my mentor, we must abide by the company’s business facts and choices, and respect the company’s position in the industry and the current status of its products; research problems must have a way to be implemented into products, rather than solving “human problems.” ".
It makes no difference to me whether I work at home or abroad, but my girlfriend just graduated with a doctorate this year. If I go abroad in 2019, it will change from a foreign place to a foreign country. When we are in different places, we can still see each other once a month, but when we are in another country, we can't meet so frequently, and there will even be jet lag, which is even more troublesome. So I quickly ruled out the option of working abroad. At the beginning of 2018, I had a sudden idea and signed up to take the TOEFL test naked, and scored 103 points. Speaking and writing were the worst, with scores of 22 and 24 respectively. With certain preparations, I might be able to improve. But then I didn’t plan to work abroad, so I didn’t deliberately learn English.
Interviews at large companies mainly consist of answering questions, talking about research results, and chatting.
From a comprehensive perspective, only large domestic companies or unicorns are left. Large companies generally recruit through several channels, including invitations from big bosses, internal referrals, or overseas submission of resumes. Because we are in the MSRA circle, many big guys from MSRA have naturally become our important choices, and they will also take the initiative to contact us. This method of invitation from big guys has the highest probability of getting top offers; internal referrals are Recommended by friends, my friend Cong, who is a doctor in Lianpei, helped me recommend several companies; overseas investment resumes are generally not recommended. Firstly, the probability of the resume being "picked up" is not high, and secondly, it is matched. The department may not be suitable either. The method of invitation or internal recommendation by a boss is equivalent to using the endorsement of the recommender, which gives the interviewer a good first impression before the interview. However, you must find reliable friends when making internal recommendations. Someone once recommended a candidate to me and said bad things about them, so I said, don’t recommend them. It’s never a good thing to say bad things about others because of personal grudges.
I interviewed a total of 12 companies, divided into three categories:
Major domestic manufacturers, including Alibaba, Tencent, ByteDance, Meituan, and Huawei
Domestic unicorns, including Pony.ai, 4Paradigm, Cambrian, and Horizon
Foreign companies, including MSRA, Xilinx, VMWare
The interview formats of most companies are similar. The first two rounds are technical. Most companies will ask you to talk about the research you have done and do some algorithm questions. Later there are interviews with supervisors, more senior CEOs, and sometimes HR interviews. These high-level interviews are mainly chat-based, mainly looking at the influence of research results, future career plans, and the degree of compatibility with the company's values. Senior leaders will also take this opportunity to promote the company and attract candidates. Some senior leaders will also ask to write code during interviews.
For example, Alistar includes two rounds of technical interviews, a supervisor interview, a cross-over interview, and a final round of Alistar interviews. The final interview at Alibaba Star was with the P11 president of the department, two big guys from other departments, and the senior HR director. They were asked to give an academic report for 30 minutes first, and then the big guys asked questions for 30 minutes. At that time, I was talking about the batch-stream integrated processing database project I was working on. I didn't realize that the P11 boss was a senior expert in databases, so I was scolded. However, many of my immature projects used the opportunity of reporting to the boss to receive feedback. When I was interning at MSRA, one time (not for an interview) Turing Award winner Butler Lampson came to MSRA for a visit. I told him about the idea of total order message transmission, and he gave very valuable feedback. This job was rejected after After 4 times, it was finally published on SIGCOMM '21.
I was also impressed by Ali’s cross-interview. The interviewer for the cross-interview was from another department. He was an expert who had switched jobs from Intel. He asked me a lot about CPU architecture. He said that I knew more calculations among my peers, but I still knew less about the details. . I said that it is difficult to understand the details of the CPU microarchitecture through public channels. He said quite profoundly that this is the value of work experience. This remark made me realize that I must join a team with core technologies and have as broad a technical vision as possible in order to learn as much know-how as possible in my short youth.
When interviewing Mr. Guo Chuanxiong from ByteDance, Dr. Guo gave me a math question. That question was to find the upper and lower bounds given a recursive formula, which was derived from his DCell paper (the number of servers increases exponentially with the DCell level), but I did not read this paper carefully. In fact, it is not difficult to find the upper and lower bounds in mathematics, but I did not work it out at the time. It suddenly became clear to me after Dr. Guo explained it to me. Dr. Guo said that the kind of research he likes is the type of research that has elegant mathematical theory and practical value at the same time, but such research is hard to come by. When interviewing Byte CTO Yang Zhenyuan, he asked me to write code on the spot and talked about a lot of details about my research work. He said that he hadn't talked about such low-level things for a long time. He was doing this kind of low-level optimization when he was at Baidu. It felt like he had been a general for several years and returned to the knights to learn swordsmanship.
When I was interviewing with Pony.ai, I met the legendary leader Lou as I wished. Before meeting Master Lou, there were two rounds of coding interviews. Each interview had two or three not difficult algorithm questions, which were very similar to interviews at Google. Master Lou gave me a few puzzles, none of which I had ever seen before:
1. Use a coin that has a probability of turning heads every time it is thrown, allowing multiple throws to realize an equal-probability 01 random variable, requiring the expectation of reducing the number of throws as much as possible. This is indeed a very beautiful question. I figured out the optimal strategy on the spot but failed to calculate the expectation. Later I found that the expectation is 1/H(p), where H is the entropy function.
2. The number of vertices and faces of an n-dimensional cube.
3. Given any starting point and end point on the six faces of the three-dimensional cube, find the shortest distance for the insect to crawl along the face.
4. Divide one side of a rectangular paper into three and five equal parts. The leader gave me a piece of A4 paper to fold. I struggled for a long time and came up with a method. The leader thought it was right, but I found out that I was wrong. Then I changed it and found a third-class method that he had never seen before. The dividing method is relatively complicated. In fact, the leader's method is very simple and can be easily extended to any equal division.
At that time, I also asked the leader Lou what he thought of the end-to-end autonomous driving pipeline. He said that life-critical things were completely left to black boxes like neural networks, and he was worried; the interpretability and debuggability of deep learning are not strong. Even if the effect is good, it will be difficult to explain to the public if something goes wrong, and it may not meet the requirements of regulations.
MSRA may have the largest number of interview rounds, including two rounds of code interviews, collective interviews with researchers and supervisors, cross-sectional interviews, deputy director interviews, HR interviews, etc. Each round of code interviews consists of two algorithm questions and one system design question. It is known as the "one-vote veto system" because those who fail to pass the coding test will not be hired no matter how much the supervisor intercedes. MSRA's group interview of researchers and supervisors is similar to Huawei's group interview of researchers and supervisors, and the final interview on Alistar. They both give academic reports first, and then ask questions and exchanges. Intersection mainly tests the breadth of knowledge and openness of thinking. There was once a candidate who aggressively attacked the interviewer's research field during the interview, which was not appropriate.
When interviewing with Huawei, Tan Bo asked me what I thought about the application of FPGA in data centers. If you just talk about the advantages of FPGA from the perspective of defending your own research results, this answer will be narrow. Different companies have different choices, including technical path dependence and commercial considerations. The company also hopes that candidates can objectively analyze the advantages and disadvantages of various smart network card architectures. When I am an interviewer myself, I also hope that candidates can break away from their own research work and have new independent thinking. It is best to sublimate it to a higher level, instead of reciting reports made a few years ago like a repeater. Say it again.
In fact, I didn’t prepare much for these interviews, because I think the algorithm questions from big companies are not difficult (unless they are math or brain teasers); the research results have been talked about many times, and there is no need for PPT materials. You can tell it; there is no need to prepare for chatting. My career plan is to become a system architect. I also like to struggle and match the values of major domestic companies. At the same time, I also maintain a strong interest in computers and hope to continue to explore Frontiers of Pasteur's Quadrant. Some interviewers will still ask some questions related to basic computer knowledge, and I just know what I know and what I don’t know. For example, I have never used C++ and Java, so I said no. Although I have done some kernel development, they are the simplest kernel modules and I have never studied the network protocol stack. When I am an interviewer myself, I also like to examine the candidate’s basic computer knowledge, such as operating systems, networks, distributed systems, databases, etc., as well as the syntax and semantics of the candidate’s usual programming language and knowledge of compilation and link runtime (such as garbage collection). ), it is often found that many candidates can answer theoretical questions very clearly, but they cannot answer the next level of questions because they have not really used these systems and only memorized some interview experience. Some candidates even try to deceive the interviewer, which is easily discovered if the interviewer is an expert in this field.
It wasn't until I became an interviewer myself today that I discovered that, just like when I was a candidate, the closer the candidate's level and experience are to the interviewer, the more comfortable the interview will be. Meeting experts is like chatting with friends. For example, if you meet someone who is doing ACM, let’s talk about what data structures and algorithms you are good at, which questions are pitfalls, and how to cooperate during the competition; if you meet someone who is struggling with the website, let’s talk about what frameworks are used. , how many pitfalls are there in CSS and JS, and what is the history of blood and tears in the operation and maintenance process; when I meet Mr. Zhihui, who is tossing about embedded systems, let’s talk about which microcontrollers have been played (burned out), and how to adjust the parameters of PID control; When I meet someone who is engaged in blockchain, I will talk about the consensus mechanism and smart contracts, and ask him to give me popular science about currency prices that I have not seen for many years; when I meet someone who has played with information security, I will ask him to tell me which websites have been hacked and found. Even if you encounter an area that you don’t understand, you can still humbly ask for advice and listen to the candidate tell his story, and you can learn a lot.
The choice of offers from major manufacturers
Thanks to the love of all interviewers and leaders, I got good offers from these companies. Offers from major domestic manufacturers will match each other. For example, after I got the offer from Alibaba Star, several other major manufacturers will refer to this package. Therefore, if juniors and juniors want to find a job in a big factory, they must not just work for one company, otherwise they will wait to be ripped off.
In order to select an offer, I wrote a summary for each company just like reviewing a paper, listing the Strengths and Weaknesses, and put the information collected during the interview process and various aspects into Comments to inside the Author (comment). At the same time, I will also record the bosses, direct supervisors, new outstanding employees and interns in the company. Just like the review comments for each paper at the top conference were very substantial, I also wrote thousands of words of reviews for each company. Based on this, I created a spreadsheet to score each company on multiple aspects, including:
Entry salary
long term expected salary
Offer's relative rank among peers
Business fit
growing space
Industrial influence
academic influence
Work-life balance (work ease)
Leadership familiarity
job stability
Company Culture
Company brand
Department Outlook
Field prospects
Technology accumulation
Boss Daniel
Daniel
Work place/residence
Each item is scored from 1 to 5 like a paper review, and then statistical indicators such as arithmetic mean, geometric mean, and variance are calculated. Based on these statistical indicators, the companies are then sorted. Just like reviewing a paper, the failure of the paper does not depend solely on the score, but in the first round of screening, those with low scores can be filtered first. There are 7 companies with an average score of 3 or above, which means 5 companies were filtered out in the first round. Because only one company can be selected in the end, and the "acceptance rate" is only 8%. This is indeed a very cruel selection. For companies that were screened out in the first round, in order to avoid hiring HR and wasting the sincerity of the supervisor, I spoke up earlier. Just like telling the author early after the paper is rejected in the first round, the author can switch to other conferences earlier. I am not good at rejection. I have to think carefully for a long time before rejecting every company.
The average score of these 7 companies is as low as 3.78 points and as high as 4.11 points. Among them, the average score of the top 6 companies is at least 3.94 points, so it is really difficult to distinguish them. I also have trouble weighting these above metrics. When submitting a paper, the final acceptance is not determined based on the score, but after discussion at the PC (Program Committee) meeting. If there is a champion (support) and no one objects, the article will most likely be successful. The same is true in the context of choosing a company. If there are obvious advantages and no obvious disadvantages, it is a better choice.
In the end, because my doctoral supervisor was at Huawei, and Huawei has the advantage of combining software and hardware design, I made a difficult choice. It is even harder to reject the other 6 of these 7 companies, because each company has broad prospects, generous remuneration, ample room for growth, and a large number of talented bosses and classmates. At that time, maybe it didn't make a big difference which company you chose.
Not long before graduation, I encountered the "516" US sanctions incident. People were panicked for a while. Many people said that Huawei was dying and advised me not to go to Huawei. A teacher said to me that the most precious thing for people is time. If you want to support Huawei, you can spend money. If you buy a mobile phone, you will support Huawei. But if you choose to join a company and invest most of your time, it's different. Qian Xuesen and other great scientists can give up their generous salary and return to China to do unknown work, but they can no longer do world-leading research and win Nobel Prizes like Yang Zhenning, and their living conditions will not be very good throughout their lives. However, after on-site research and interviews, I confirmed that Huawei is not as vulnerable as outside rumors. The entire company is still running normally. What projects should be done are still being done. While eating in the cafeteria, the TV series "Sanctions and Sanctions" are being broadcast. Relevant news, and everyone has become accustomed to it. Therefore, I insisted against all opinions and insisted that joining Huawei and Qian Xuesen returning to China were completely different in nature, and I maintained my choice. It was also the major choice I faced the most opposition in my life.
During the onboarding training for new employees, we had a discussion with a president of the human resources department, and I proposed whether we could follow the example of "Alibaba Star" and launch a systematic project to attract outstanding talents. In August 2019, the company happened to launch the "Genius Youth" program, and I was honored to be a member of it. The company-level "genius boy" program certainly has nothing to do with my casual suggestions, but I can predict that companies are paying more and more attention to the recruitment of outstanding talents such as Ph.D.s, rather than being punished as some people imagine. Recruiting outstanding talents. Three years have passed, and with the care of my leaders and colleagues, I have assumed increasingly important project responsibilities and exerted greater and greater value, which also proves that the choice I made back then was not wrong.
Although I only have very limited work experience, what I want to tell my fellow students is that among the above-mentioned "scoring items", leadership and departments are more important than the overall situation of the company and research field. Whether a person lives comfortably and whether he or she can be taken seriously at work largely depends on the leader, especially the direct supervisor. If you are consistent with the leader's vision, cooperation will be smoother. It's like in the process of studying for a Ph.D., the mentor is far more important than the reputation of the school and college. Secondly, the importance of the department depends on the business and atmosphere of the department, that is, whether the fundamentals of the business are stable, whether there is room for imagination, and whether the team atmosphere is harmonious. In large companies, a decline in the company's stock price may affect income, but the short-term impact on individual work is not significant. The popularity of a research field is a slow process of change. If one day your research field becomes obsolete, it is not too late to change it. Many big names have experienced switching research fields.
Good stuff to learn, like three times in a row ↓