I just received a summer internship offer from Tencent, which means that my interview journey is over. This article is used to record my short (more than 20 days, only three technical interviews) but very fulfilling interview journey. (Mianjing and chicken soup are attached at the end of the article)
1. Born in a background
of majors, Shuangfeiben + scumbag, no papers, no related projects, no Top competition experience, such a three-no contestant has embarked on the algorithm road of Ragnarok .
The direction within the group has nothing to do with the field of artificial intelligence, so it is usually a self-study. The personal skill stack is roughly: machine learning, recommendation algorithm, NLP (partial classification model only), CV (partial classification model only), reinforcement learning (not reviewed, forgotten), in fact, you can get a general understanding of the blog I wrote. The range of my knowledge reserve. Not too much, they were all learned in the postgraduate year (100% self-taught), and the undergraduates were decadent. (You can compare your skill stack and find that you can actually enter Tencent)
When I was researching, I would learn algorithms in every direction. Later, after I learned about the scene of fights between gods in some directions (such as CV), I turned to recommendation algorithms and NLP. In addition to gnawing on the principles of algorithms, I will contact some data mining competitions to test my understanding and practical ability of algorithms. There are no teammates, no machines, and no top results.
While I usually study theory, I also pay more attention to practice, and try to reproduce the algorithms I have learned with code. It is really important to be able to implement the algorithm. No matter how thoroughly you eat the theory, you will not be able to use it in practice. Everything is in vain. Therefore, it is recommended to write more code and reproduce the model as much as possible.
The above is the landlord's learning process, which is bland and unremarkable, so the resume has no bright spots, so that even some companies can't pass the resume screening.
2 Interview process
I prepared relatively late for the interview. I didn't want to study at home during the Chinese New Year, and I didn't officially start preparing for the interview until the start of school in March. (It is strongly recommended that you prepare early. Big factories will start recruiting during the Chinese New Year. The earlier the delivery, the greater the advantage. As long as you reach the qualified line, there will be a pit for you. , that's really a fight between gods).
I started to focus on writing questions, summarizing what I learned, and making a resume in early March. Before I was fully prepared, I never dared to invest. After 3.20, I invested in some companies one after another, about seven or eight. There are only two types of positions for recommendation algorithm and NLP algorithm engineer. (I don't know how to develop, so I didn't leave a way for myself)
So far, most companies have just passed the computer test (Huawei, Baidu, Ali, NetEase), and interviews have not yet been arranged. Meituan, Tencent, and Meituan died on the one hand. Tencent successfully completed the three rounds within half a month, and finally completed the final HR round on the last day of advance approval. (See below for the detailed time node and the content of the face.)
3 face to face
Meituan
3.19 internal recommendation algorithm base Shanghai
Machine test: (3.27) One
side: (3.31) 40m
self-introduction
digging project
The project mentioned loss weighting and asked me if I had considered using adaptive weight
attention and self attention to distinguish between
transformer
word2vec
cbow and skip-ngram, application scenarios Programming question : tear off the
01 backpack by hand
, career planning within three years,
which companies have you invested in, and where is the base? But a backhand hang makes me doubt my life)
Tencent
3.24 pushes machine learning base Shenzhen
Machine test: (4.4) One
side: (4.2) 50m
self-introduction,
deep digging project ,
confrontation training
, specific methods of multi-task training (ask if you have tried another method, but I don’t understand)
feature engineering
xgb and lgb
. Do you consider models of different natures? Fusion between
Probability questions: flip a coin, whoever flips the coin first wins, asks the winning probability of the first flip (from the perspective of probability and law)
Programming problem: common prefix of string arrays
Two sides: (4.8) 35m
self-introduction,
digging deep into the project ,
asking about the difference between lstm and gru in the research direction of the group
(I mentioned the complexity, and asked how much the difference is)
The difference between
xgb and gbdt The difference between xgb and lgb
The shortcomings of lgb ( I haven't noticed the shortcomings)
C++ virtual functions (not)
deep copy, shallow copy
Know which dimensionality reduction algorithms
Do you understand spark, hive (do not understand)
internship time, graduation time, employment or further study
Rhetorical question
(the second side is over, the official website process is directly here) hr side, one less round of technical side)
HR: (4.14) 25m
Self-introduction
Project difficulties and solutions (is there any new solutions in the future), highlights
Why undergraduate students choose computer science Why
graduate students choose this direction Do you
have relatives in Tencent
Your own strengths, strengths
and
weaknesses, the impact of your weaknesses on yourself, How to correct shortcomings
Other people's evaluation of yourself
Company invested and offer situation
Family situation
Family's tendency to base
Growth situation from childhood to adulthood
Rhetorical questions
Telephone oc (4.15)
official offer (4.16)
In fact, the interview is not as difficult as you think, it is a very basic question, so the big factory is not as far away as you think.
The following should be your most concerned questions...
Is Algorithm Gang really gone? Can't go ashore without a summit meeting? Following my example, you will know the answer is no.
The algorithm post does have a lot of competition, but it is not as serious as the friends described. In fact, most of the answers to this question should have been written by those friends who failed in the algorithm post, and some exaggerated elements are mixed in it, so 70% of the credit is enough.
Every company's recruitment does mention the thesis requirements, but no company puts the requirements in the necessary skills, and they all appear in the form of bonus points, so the top meeting is only a sufficient but unnecessary condition.
If you're really interested in algorithms and think you can go a long way in this field, go all the way. (If education is not an advantage, it is best to stay behind and learn some development skills. Although large factories do not value education so much, some small and medium-sized factories have very strict qualifications.)
Follow up:
Byte fishing, Meituan fishing twice, Huawei appointment, NetEase appointment, all refused. So after only three technical interviews, this is the end of the internship, and prepare to vote again for the autumn recruitment.
4 Chicken Soup
You, who are reading this article, must be under pressure and unable to extricate themselves
I also thought about how great it would be if I could return the notice and forget about the cold window of the past ten years.
But after thinking about it, if I can really go through this again, then I must work harder and not let myself be troubled by my lack of ability.
So if you are now, the road to work is bumpy, there is no need to put too much pressure on yourself. There is more than one chance. If the spring recruits are not smooth, what about the autumn recruits? A few months of hard work is enough to make you reborn. I have seen too many autumn recruits that did not go well, and spring recruits took the example of a large factory sp.
So it's never too late to start working hard, even now, all your hard work will pay off at some point in the future. This moment may be far away, but it may also be very near, and an offer may come inadvertently.
Brothers, I wish you all the best of luck!