Write in front:
I am a scuba diving party. I have been secretly observing Niuke.com for a long time, and I have also obtained a lot of useful information and help. I have been planning to write a final article to give back to the cattle. Recently, I have been busy with the autumn recruitment papers. Then just get busy playing. It has been delayed until now, and finally I have calmed down and completed this sharing before the arrival of 2018. The content is not only a face-to-face scripture, but also a summary of the various things that have been done to find a job this year. Perhaps the experience and experience are far from those of the great gods. But I also hope that it will be helpful to the Niu Ke Niu friends and the teachers and brothers in the school~
LZ background:
Briefly describe your personal situation. Double ordinary 985, has been going to school in his hometown. Undergraduate Computer Science and Technology. During the undergraduate period, I was more playful, with average grades and mediocre hands-on ability. So when I was a junior, I didn't have the courage to participate in the autumn recruiting test. Just get ready for the exam. Postgraduate entrance examination is also convenient for direct examination of the school. Graduate student majoring in software engineering. During the master's period, the nature of playfulness has not changed. After playing nearly 1400 dota2 rounds, when the Legion came back, he even saved a wave of Azeroth. But I know more about time management than before. The proportional control between play and learning is also well done. I've been trying to figure out a way to improve my skills. An obvious difference is that the big homework during the undergraduate period is all mixed with the thigh of the roommate. The projects during the postgraduate period are basically done by myself and my classmates.
Because the undergraduate thesis is the content of the machine learning direction. Since then, graduate students have also been in charge of projects in this direction. So choosing this position to find a job is a natural fit. We are lucky, because the direction of our ordinary people's undergraduate graduation thesis is randomly assigned by the college.
In general, the postgraduate years have indeed improved a lot. On the one hand, it comes from the project in charge of the laboratory, and on the other hand, it is to think of various ways to improve it. The biggest meaning to me during the master's degree is to buy myself time to improve. For the first study, take classes + complete large homework + supplement the basic knowledge of algorithms and mathematics by yourself. In the second year of research, the laboratory is a teacher's project + competition + job search. The third year of research is, of course, the completion of the graduation thesis, plus preparation for future work.
To sum up, I know how to combine work and rest, and I have a better mentality. I can create a good mood for myself to study, and I have more perseverance for goals.
I will write the following sharing according to the timeline after the Chinese New Year in 2017. It is also convenient for me to remember~
Spring recruiting preparation:
After the end of the year, start preparing for the internship. Of course, learning needs to be continued, and this period is equivalent to summarizing and assaulting all aspects. In fact, it is mainly these aspects when studying. The goal of the internship job is to be a machine learning algorithm engineer. The content of preparation is divided into the following aspects:
1. Review and summarize previous projects. Including the projects that the laboratory is responsible for, and the parts worth talking about in the projects you do yourself. At least you need to be very familiar with your project and have to withstand the inspection of the interview. The interviewer may question the project from various angles.
2. Online OJ brush questions. The time before lunch in the lab was mostly used for writing questions. Mainly leetcode, and lintcode. The difficulty of the questions is mainly easy and medium, with a small amount of hard. The language is mainly python, with a small amount of java. According to the requirements of algorithm engineers, at least one of Java and Cpp is required, and python is also required. I personally feel that the company's requirement for interns is "knowing", not "proficient".
3. Preparation of basic knowledge of algorithms. The great gods in various reference books and information discussion areas have summarized a lot. Simply put, it is "Statistical Learning Method", Teacher Zhou Zhihua's watermelon book, etc. We will discuss with those who need it. It has the ability to push down the classical algorithm. Some algorithms have the ability to implement their own handwriting. This can refer to "Machine Learning in Practice" and "Collective Intelligence Programming" (should be the name...)
4. Machine learning package, understanding and use of various frameworks. Here I recommend a great god to everyone, which is very helpful for getting started with various frameworks. Mofan: https://morvanzhou.github.io/ . He also has his own channel on Youku.
5. Prepare a resume. There are also various strategies for how to write a resume, so I won't go into details. Then find a place to take a good-looking ID photo. This will work better. For example, my ID photo was evaluated by my girlfriend and lost 50 pounds. . .
6. Choose your own target company. If you don't have a goal to invest in the sea, it will easily lead to lack of energy.
Well, I'll add if I remember something.
Spring recruit
Having said so much, in fact, I only voted for one spring recruit. It's Alibaba that was pushed by a buddy who was an undergraduate. With him in the same department, belonging to the Ali mobile business group. The position is algorithm engineer, machine learning. In fact, it's not that I don't vote for others, there are some reasons. . . For example, at Tencent, I did not find machine learning in the recruitment list, and it was not until it expired that I found that machine learning was in basic research. As for Baidu, because our teachers have some opinions on Baidu, if we go for an internship, we are afraid that it will be troublesome to consult with teachers during the summer internship, so we do not vote. And then there's that, the internships introduced by the seniors try to go to big factories as much as possible. So in the end only voted for one. Briefly describe your interview experience.
It was successfully pushed around March 1. One week after submitting the resume, the hr in Hangzhou called to arrange an interview time.
one side
Technical, phone.
It's been a long time, so I can't remember the specific questions of the interview very clearly. Can recall roughly the following:
1. Self-introduction
2. Discuss the projects that the laboratory is responsible for. Very detailed. From the problem solved by the project, why do it, what is the meaning, what technology and algorithm are used, etc.
3. Ask what algorithms you are familiar with. Let me introduce it and discuss it until I can't answer it.
4. Neural networks are discussed. Especially BP neural network. Details of backpropagation etc.
5. For a specific scenario problem, how to use machine learning technology to solve this problem.
6. Do you have any questions for me?
That's about it. About 50 minutes or so. It was also the first time I did such a long phone interview. Relatively nervous, but the interviewer will guide me to answer questions. Questions that cannot be answered will also be discussed from a different angle, and prompts will also be given. The tension is also removed at the back.
two sides
The day after the first meeting, HR called to arrange the second meeting. Because the lab computer doesn't have a camera, I still made an appointment for a phone interview. On the second side there are probably the following:
1. Self-introduction
2. KNN is discussed. Discuss issues about dimensional explosion.
3. SVMs are discussed. SVM's overthrow, principle, and optimization method. Why SVM can solve classification problems.
4. Algorithm questions. Since it's a phone interview and can't write code, I think about it myself, and then discuss ideas.
About 45 minutes or so. Compared with the one-sided interviewer, the second-sided interviewer (should be the team leader) puts more pressure on it. On the one hand, the performance is not very good.
HR side
After the second session, there was a long wait. It was so long that I felt that I must have hung up and prepared to review and prepare for the autumn move. As a result, I suddenly received a call from HR to arrange an interview at the end of April. And then the HR interview the next day. I heard that Ali's HR is also very important, so I was very nervous and prepared some questions.
The final interview was relatively easy. Just like chatting. The questions are some of the usual HR interview questions. He also allowed himself to evaluate his previous technical performance.
About a week or two after the HR interview, the HR of the department called to communicate the benefits during the internship and confirmed the letter of intent for the internship.
Participate in the competition
Participated in the JDATA data contest in April. Since I was waiting for news after the interview, I didn't know if I could go for an internship during the summer. I also felt that my resume was not rich enough, so I participated in this competition held by Jingdong. When forming a team, because I wanted to exercise more by myself, I didn’t find teammates in terms of algorithms. The girlfriend and the other teammate are both in charge of feature work, mainly thinking of new features. I am responsible for data processing, feature extraction, data set construction, model training and parameter tuning.
The final result is fairly average. A few places away from the Excellence Award. But throughout the game, I still felt a lot of improvement. Including a lot of experience in features, models and algorithms, and parameter tuning. From the beginning of the competition, I stumbled, and every day when I submit, I rely on the sky to eat. Later, I wrote functions to extract new features every day, added them to the data set, and trained new data to submit. Every day grades progress is predictable and you know why. In the last few days, the results have also been rising. Due to the lack of time and a person's limited energy after all, the results have stopped there.
There are many competitions now, and students who are energetic can try it out. Grades are only one aspect, and the exercise you get from it can be very rewarding.
Summer Internship
I spent the entire summer as a summer internship in Ali. The specific experience during the internship I will not write so detailed, interested students can ask me.
It's actually not quite what I imagined. Algorithmic interns do not have access to production environments due to data confidentiality and security requirements. During the whole internship, I was mainly responsible for understanding the advertising industry, reading papers in related industries, doing some experiments on algorithms and strategies, and also responsible for some research work. The main improvement and exercise is to see how algorithm engineers work in large companies, and to learn from colleagues around them. Colleagues around me, especially my mentor, are very nice people with strong skills. Internship around the great gods is to discover and bridge the gap. Of course, colleagues and leaders also take good care of the interns. The transition from the campus to the work environment is huge, and the internship gave us this opportunity to get in touch and adapt ahead of time. When I go back to school after my internship, I still have time to work on my shortcomings.
The process of changing the internship to a regular began in August. On the positive side are the leaders of most of our departments, mainly introducing the work during the internship. Then there is the chat. Mainly, the boss introduced me to the future goals and development of the intelligent marketing platform. In the back, the boss is mainly talking about me listening... The second side is Ali Mom's boss, and he also talked about the work during the internship. Then there is the chat. We talked about Taobao and Alimama. The general atmosphere is also relatively relaxed. Then there are the three sides of Miss HR, still chatting. After the internship turned into a full-time interview, there was no tension and distance from the phone interview, and the chat was more enjoyable.
The final result is also very happy. On September 3, I got the letter of intent for the autumn recruitment. On September 6th, he resigned and returned to school. To be honest, this is also one of the best results I can look forward to when I look forward to my work. Since the group stipulates that the offer will be issued only after the three parties have submitted it, and the college has issued the offer relatively late for some reasons, so the official offer was not confirmed until the end of November.
After returning to school, I gave up the feeling that the autumn recruitment was over before it started~ I still participated in the autumn recruitment and invested in some companies. One is because I was looking for a job with my girlfriend, and I struggled to at least go to work in the same city in the future. The other is that all the classmates and friends around are looking for jobs. The lab's projects have largely stalled. Idle is idle. Also, if you are idle in the lab, you will look like a different kind... Let me introduce the interview experience of Qiu Zhao according to the time.
Mogujie Algorithm Engineer-Machine Learning offer
technical side
An interviewer for the Mogujie advertising algorithm team. First, I introduced myself, talked about my understanding of the computational advertising industry, and briefly introduced my work during the internship. The related issues of LR are discussed, which are discussed in detail in this part. Then regularization related issues are discussed.
The whole time is about half an hour.
technical side
Should also be an interviewer on the same team. A few questions were asked about the items on the resume. Some issues regarding the calibration of classification results are discussed. Discussed that classic Facebook article on computational advertising. The same time is about half an hour
Three CTO side, video side
The CTO of Mushroom Street. At first I thought I was going to write code online or something. The result was a discussion of some of the items on the resume. Then there is the chat. Let me ask some interesting questions and then discuss. Probably less than twenty minutes.
Baidu Algorithm Engineer - Machine Learning
The first time I went to run a job fair in Xi'an for an interview. Phoenix Nest's interviewer. Because I just returned to Xi'an from Beijing at that time. Did not prepare for the interview. Being abused by the interviewer was terrible. Every question was discussed until it could not be answered.
Probably the problem is: the understanding of the advertising system. Loss function for SVM/LR, derivation. Live handwritten code. Ask whatever is written on your resume.
At that time, it was really not reviewed yet, and the interviewer saw it. It didn't perform well anyway.
Tencent Technology Research - Machine Learning
a phone face
IEG's phone face. It was CDG that was approved in advance, and the resume was picked up by IEG for an interview.
The interview questions include: questions about the project on the resume, simple discussion of the internship, detailed discussion of LR, regularization problems, parameter adjustment methods, discussion of XGboost, understanding of game AI, and understanding of AlphaGo.
two sides
The second side is the on-site interview, and the interview department has become the department of MIG Tencent Maps. I still don't know what's going on here...
Interview questions: Questions about the project on the resume, discussion of LR, derivation of SVM, handwritten code: The first one is about traversing the binary tree, no stack, no recursion. The second is the local inversion of the linked list. Then the problems related to the map algorithm are discussed.
Due to the temporary change of department and insufficient preparation, the map algorithm basically does not understand. Handwritten code is also average. In the end it didn't.
China Merchants Bank artificial intelligence research and development offer
China Merchants Bank Artificial Intelligence R&D Department's special job fair at the school. I went to the site to submit my resume.
face to face
One side is at the Xi'an branch of China Merchants Bank. three interviewers. Mostly discussions are about the CV and the items on the CV. Then it mainly discusses Xgboost, including advantages, loss functions, optimization methods, etc. Then it's mostly up to me to ask questions about what I want to know.
Live video on both sides
The second side is also in the Xi'an branch. Video interview with Shenzhen Head Office. There were supposed to be three interviewers, but when it was my turn, only one was left. It just feels sloppy. I can't remember the interview questions, just a simple chat about regular questions.
When the offer was finally issued, the head of the head office came directly with the offer. Call the seven people admitted by Jiaotong University to the branch. If you sign it, you must sign it on the same day, and the penalty is 10,000 yuan. It's very stressful. In the end, only one of the seven people should sign. Everyone else is gone. I called again later, but didn't come. Another point is that in the process of recruitment, it has always been emphasized to us that the package is specially approved for this position, and it was never given by China Merchants Bank. But later I heard about an offer that is much bigger than this package. It may be that the salary of the branch is not the same as that of the head office. In addition, the situation such as monthly salary is relatively secretive, which may be the practice of traditional banks.
JD.com Algorithm Engineer-Machine Learning offer
technical side
On the one hand, the interviewer felt very young. Chatting is easier. Talk about school projects. Loss function for LR, derivation, pros and cons. Experience in machine learning. Jingdong's game. Finally someone talked about this game. The experience of tuning parameters, the role of regularization. Finally, their departments and work are introduced.
technical side
The second side should be the leader of the department. In the first fifteen minutes of the interview, I talked about the projects I had done and the competition of JD.com. Then start chatting. We talked a lot, and it was also the happiest interview during my entire autumn recruitment period. It was also very pleasant to chat.
Three HR faces
normal HR face.
JD.com is also one of the rejected offers that the interviewer feels the most sorry for. The first is to have a good chat with the interviewer during the interview. And in the follow-up communication, the interviewer also gave enough patience and respect. And after the offer was sent, I learned that I was not satisfied with the salary and I also increased my salary. At that time, I also wrote a guilt-filled email when I replied.
Today's Toutiao Algorithm Engineer - Machine Learning
The face is relatively sloppy, and there is not much preparation, and it is also a domineering face. Mainly to accompany girls and classmates.
The interviewer is making content recommendations. He is not interested in the content of the advertisement part, and I don't understand some of the recommended practices in the content, so I can't chat together. Then came the handwritten algorithm, talked about XGboost, and talked about some cases recommended by the content during the use of Toutiao.
There is no suspense hanging.
WeBank data mining offer
WeBank came relatively late. It was over in the hotel. I originally went back to eat hot pot with my sister and classmates. As a result, after eating, I called and called back to the second side. Been facing for a day.
The content of the interview is to have an in-depth discussion on a question during the written test. It's a question about store sales.
The second interview was the morning interviewer. Several candidates sat and chatted with him. Talked a lot. Chat from all over the world. Then call for an interview with the head of the department. The boss of the department is dissatisfied because our team does not know enough about the micro public.
Finally invited us to dinner~ I left first because my sister was waiting.
Then the HR interview was carried out as normal.
One of the reasons for not going in the end is because the content of the work is relatively simple, and it may lose competitiveness. The other is that the money is given less. It may be because the data mining is not the problem of algorithm engineers.
Finding a job summary
Let’s summarize some of the experiences during the job search:
1. Resume must be done carefully. This is the entry point for all interviewers to know you. Don't fake your resume. Many interviewers will ask you one by one according to what is written on your resume.
2. Practice as much as possible. Experience the state of work in advance. Normalization is also relatively straightforward and simpler. So plan ahead and start preparing ahead of time. The recruitment battle was officially brought forward to the spring of the first year. In addition, the internship subsidy can also save a little money for graduation trips.
3. Push inwards as much as possible. This is obvious, many internal pushes can still be exempted from the written test.
4. The foundation should be solid. Whether it is a major or a major, the foundation is very important. Includes basic data structures and algorithms. Classic algorithms for machine learning.
5. One or several proficient languages. In a word with hippo, language is the weapon for programmers to walk the rivers and lakes. If you don't even have a weapon, how can you walk.
6. Go to Niuke.com to see the classics ~ see the experience of the predecessors.
7. During the interview, I think fate or luck accounts for a large part. It is not that you are looking forward to entering a certain company, prepare carefully, and you will have good results in the interview. Perhaps the last received was an inadvertent offer. So don't set too strict goals for yourself before the interview. The target should be a series of companies, and when facing the companies in it, they will go all out. The final result depends on fate.
8. Have a good attitude and don't be discouraged if you fail. Summarize more, just like summarizing wrong questions in middle school. There are many companies and many opportunities to try.
My 2017 summary
Construction will be done here today, and there will be time to continue construction later~
ps. If you have any questions, you can leave a message or private message me, or contact me on WeChat baihao8904. After reading this for a long time, I hope I can help you solve some problems.
If you reprint, please contact in advance and indicate the source~
Author: Big Wild Goose Pagoda at 4am
This article comes from Niuke.com
——————————
Niuke.com (www.nowcoder.com)
- Real Questions for Written Exam of Internet Famous Enterprises
- On-campus job-hunting pen & face
- Programmer job placement information
- Programmer learning exchange community