Almost all natural language (NLP) positions require 3 or more rounds of interviews. The first round: basic communication (self-introduction, experience introduction...), basic knowledge (threads, design patterns...), basic algorithm questions (Transformer, Bert...). The second round: a detailed discussion on algorithmic questions with certain difficulty and project experience. The third round: the final interview of the technical leader and HR. And must take the arithmetic question. For every company, the algorithm is definitely a basic line. If the algorithm fails, there is almost no chance of reaching the third round of interviews. So algorithm preparation is required! But it’s not a panacea. If you think you can pass the interview just by brushing Leetcode, then you are a bit blindly optimistic. Let's start to share the dry goods, which are very dry, even dry than dried shiitake mushrooms! Everyone take the small benches.
The key to passing the first and second rounds of interviews: algorithm questions!
In each round of interviews, if there are algorithm questions, it is best to do them accurately, so that there is a high probability that you can enter the next round of interviews; if you have no idea at all, it is basically cool! If the thinking is correct and the writing is not particularly correct, then there is still a probability of about 60% to enter the next round. During the interview process, if you encounter a difficult topic, if you have no ideas, you can write a version of a more complex method first, and then ask the interviewer if he can give some hints. In the worst case, it is really not possible. Ask the interviewer if you can change the topic (mainly depends on whether the interviewer is willing to change it). If you change the topic, you must do it, otherwise it will be cool, I have done this. Most of the questions are based on Leetcode with medium difficulty, and the proportion of the original questions is relatively high, or it is a deformation of the original questions; very few hard questions, big companies will take the exam.
So here comes the problem! How to solve so many algorithm questions? How much brush is appropriate? 50 roads? 100 roads? 200 roads? Of course, the question sea tactics are always effective! But our time and energy are also limited. It's not just a matter of probability! Even if you have done 300 questions, but you have not fully understood or mastered them, then you may overturn if the questions are slightly deformed. Therefore, the strategy should be: focus on preparing for the high-frequency algorithms that various factories like to test and combine them with brushing questions. Through research, it is found that the high-frequency algorithm test points include: simpler quick sort, edit distance, binary tree, recursion, random number problem, matrix rotation, etc. There are also theoretical questions closely related to algorithms: B ptt , Lstm , Transformer , GPT and so on.
Algorithms are the core and foundation of passing the interview. If the algorithm fails, then there is no chance. So everyone must pay attention to algorithm preparation, do more questions, and be quantitative! At the same time, the key algorithms must be understood and even implemented with code. The course "Intelligent Thinking and New Knowledge-Algorithm Interview" has been comprehensively sorted out and explained. Each algorithm has been introduced and explained in detail. If you are interested, you can contact the teacher to ask for it.
The key to the third round of interviews: full preparation, good attitude and communication skills!
After the "hail of bullets" came the third round. This is the "final battle" that decides whether you will stay or not! You have to pass two levels, one is the technical team leader or technical director, and the other is HR. The former is generally the object of your future reports, and most of the questions you are asked are technical issues and project experience discussions, as well as some personal qualities that he will focus on when working with him in the future. The latter is the consideration of company culture, personal character, personal growth, and salary negotiation. Let's help you analyze the key points of preparation through the important roles in the two interviews.
Several classic questions that technical leaders like to ask:
How to evaluate an algorithm project? In addition to development work, what else is required for a complete project?
What is the main algorithm technology point used in the project in your resume? What problems were solved? How effective is the boost? What changes can be made in the future to make it better! Can the application scenarios be diversified?
How do you develop projects? How to land? How to deal with insufficient resources?
What do you consider your strengths and weaknesses? Why?
Do you have anything to ask me?
Answer suggestion:
Prepare your self-introduction carefully, highlight your own advantages, and be differentiated. It is best to be able to recite proficiently. Concisely introduce the project experience and achievements. Points are awarded for academic achievement and competition results. Be humble, honest, pay attention to hear and understand the interviewer's questions, think before answering each question, and don't be too tactful. Fully understand the job requirements of the interview company and the interview position, especially the company's business.
Several classic questions that HR likes to ask:
Your strengths and weaknesses?
What are your salary requirements?
What do you think about overtime?
How do you deal with conflicts with colleagues and superiors at work?
Sum up yourself in three words?
Answer suggestion:
Don't Versailles when talking about pros and cons! Don't mention some very serious shortcomings, such as: laziness, bad temper, etc. Salary requirements should be reasonable and not too "greedy", otherwise it is easy to lose good opportunities! Generally, the trade-off of salary and treatment should be made from the perspective of personal development. Usually asking about working overtime is actually testing your "whether you are willing to contribute to the company"! For conflicts between colleagues and bosses, the best solution is effective communication and empathy. If it doesn't work, you can also ask a third-party authority to judge. Summarize yourself with three words, mainly highlight your own advantages, and have certain differentiated characteristics. Of course, HR asks more than that, and I hope that everyone can go to the interview with a positive attitude, confident, stable, humble, and sincere.
interview strategy?
It is recommended to start with the small factory first, and then go to the factory you want to go to the most . Of course, small factories don’t need too many interviews. Generally speaking, the interview mode and content are similar, and the main purpose is to accumulate experience. Therefore, students who are more confident in algorithm questions can directly face large factories after accumulating some experience in small factories.
Interview difficulty ?
According to the feedback from many employed students, the difficulty of NLP-related positions is roughly as follows, in order of difficulty from high to low:
Tencent [ WeChat ], Byte > Huawei, Meituan , Shopee > Bilibili , Zhihu, Xiaohongshu, Netease > Weibo, Ctrip