Preface
The article is more about the record of my autumn recruitment process is relatively long, and there is not much dry goods, you can directly pull it to the section on the face.
On October 9th, Tencent called an official offer call. The job was an NLP algorithm engineer. After communicating the salary, the offer was accepted, and the autumn recruitment officially ended. Autumn recruiting results: Outside the goose factory, Meituan and JD.com, Huawei is still soaking in the spare tire pool. After considering the region and department, it basically chose the goose. In the autumn move, I told myself that after the long autumn move, I must write an article to record this journey, but after I felt sad, I wanted to say it, and I have delayed writing this article until now. After the hypocritical words, turn to the topic, first introduce myself, 985 C9 masters (guess which school, I am too south), undergraduate majors in electronic information engineering, graduate majors in biomedical signal processing, how can I be a biomedical professional Started NLP, this has to start from a year ago.
Anxiety is the driving force of instinct :
The older brothers and sisters in the laboratory can generally get the algorithm post of Dachang. A year ago, I think that step by step will be able to go back to the revolutionary road. But after communicating with a senior who was looking for a job last year, I panicked. Last year, the cruel posture of the algorithm job emerged. My first research direction was algorithmic edge ball, the second was the golden internship, and the third was the weak programming ability, which fully shows that I have been persuaded in advance. Return. But anxiety is the instinctive motivation. If there is no advantage, I can only prepare earlier than others. In a month, I struggled to finish the top 150 middle questions of the leetcode and I watched the teacher Chen Yue from Zhejiang University. "Data Structure and Algorithm Course", there is also a project in the deadline to continue to advance, and this kind of multi-line combat has become the norm for the next year.
When never ready:
Refining questions and preparing is a process of continuously increasing anxiety, because the more you learn, the more you have to learn, and you will never have a good time. So, I took the initiative. At the beginning, I was looking for an internship. It was really aimless. I didn't know who to push or where to find internship information. Niuke made a lot of online investment, and finally only one company gave an interview immediately. Respond, see if the other party is really short of people, but the face is very delicious, there is no more information; then, I saw a student post (CV's Unicorn Company) posted by a classmate for his senior, and see the comparison of the internship gold content. Gao, so I tried to cast another wave. When I was interviewed, I answered all the lucky questions. The alumni also said that I had a strong desire for internship and passed the interview. The internship after the autumn recruitment is generally easier to find. The last intern has already returned, and the formal summer internship has not yet started. Then, if you have a chance, you still have to try, because you never get ready. I’ve seen "Random Strolling Fools" before. The core concept in the book is that the world is random. If you succeed, either you are lucky, or you try to increase your probability.
Rolling snowball :
For personal reasons, I have always been inclined to develop in the Pearl River Delta, and the goose factory is naturally one of the major target factories. During the internship, I did CV-related work. The work intensity was close to the business. The laboratory had to be taken care of. Then the flu struck and I was recruited. The company laboratory and the cold were exhausted. After two months of internship, I gradually felt that the business of CV has become a red sea. It is not the algorithm capability but the business capability that determines the company's revenue. Start-up companies are relatively weak in this respect compared to established companies. By chance, I learned that my classmates have already practiced in the goose factory, and the department's NLP direction is relatively short of people, because it is the goose factory. At that time, Bert just came out and thought that NLP seemed quite promising. Now it seems naive to think about it. So there was a wave of interviews, and the other party wanted to do some work that combines CV and NLP. Seeing that I have a CV background, the interview process is relatively fixed, the internship interview is passed, and I will go for an internship after the year. Passing the second internship interview is similar to snowballing. The first internship experience plus the normal performance of the interview will pass the interview smoothly. Prepare in advance, it is easy to accumulate the advantages of the early stage, but don't waver. For example, I almost overturned.
The slack period of long-term combat :
After the winter vacation, I went back to study the questions, but did not do it; after the winter vacation, the spring recruitment internship began gradually. Several summer internships were prepared to brush the questions, but there was no question or face to face; and the autumn recruitment was planned before the start of the early approval. Make good preparations for the questions, and go to the iron water competition again. If you haven't scored yet, you haven't been ready. During this period, the laboratory and company affairs did involve some energy, but the main thing was still slack. The difficulty of the summer internship interview and the winter internship interview are not at the same level. They are very close to the difficulty of the formal interview in autumn. For example, the summer internship interview in Ali is equivalent to the selection of regular employees. I have not experienced summer internship interviews, underestimated the difficulty of autumn recruitment interviews, and there is no basic knowledge such as systematic questioning and preparation of machine learning. The subsequent autumn recruitment interviews are really tears of men and women.
Chaotic rhythm, four in a row
On the train back from the Dragon Boat Festival, I realized that I couldn’t delay any longer, so I rationalized my thinking: Goose Factory has become a daily internship, and the opportunity to become a regular employee is unknown. Even if the department’s base Shanghai becomes a regular employee, it can only be in Shanghai. Continue to interview; job selection, the competition of CV is currently fiercer than NLP, and the current internship is also related to NLP, and I finally choose to find an algorithm position in NLP, and I will work in Shenzhen. The goal is very clear. Starting to invest in overseas, it is also the top iron, and basically invest in the core departments of various major factories: Huawei Noah's Ark Lab, Toutiao AI Lab, NetEase Interactive Entertainment Artificial Intelligence Lab, Baidu Dasou, Baidu Natural Language Processing Department , Tencent AMS, JD Advertising; also invested in some such as Green Factory, Blue Factory, Huya and Bigo. In preparation, I read the NLP interviews on Niuke and sorted out some common problems. By the way, I sorted out the common problems of deep learning and machine learning. Then I followed the picture to find out and listened to the story of the brother's half-book "Sword to Offer". , I did it diligently. In addition to preparing for the autumn move, it has already become a third-line battle. The company and the laboratory have a lot of things. Basically, I can only take a look at it. In the later period, it has become an interview-oriented study.
7月中旬，面试开始扎堆出现，最开始的面试怎一个惨字了得。华为诺亚直接简历挂；百度大搜先是特色大数据体不会，又出了个剑指offer题，叫我用循环莫用迭代，磕磕碰碰写出，而后挂；绿厂比较顺利进入二面，二面不聊技术也不问项目，聊NLP的未来，向做用户画像的面试官安利智能对话的美好场景，成功说服了面试官而后挂；绿厂挂完，下午就是头条，根本不在状态，又深圳只有CV岗，硬着头皮面还是惨兮兮挂了。四连挂之后，我慌了，秋招面试太南了，本以为是谈笑风生，但都是反手一个挂。
随机的骰子， 惊险上岸
生活还是要继续，向HR朋友咨询后，我发现问题除了基础模块准备不充分之外，简历与自我介绍是个大问题：此时简历几乎无更新，项目介绍几乎都是一句话概括；自我介绍，也是姓甚名谁家哪的哪实习过的30s自我介绍，面试官此时简历都没看完呢。这两个问题导致，面试官不能在初期通过简历以及自我介绍了解你，就只能根据他的考察方式去考察你，那你就很被动了。花了两个周末的时间，我认真打磨了自己的简历和自我介绍，核心就是逻辑，自我介绍时要注重语气与节奏；同时，面试真题补了一下。
接下来的面试，总体来说是顺利但又曲折，但是life is random walk，一切充满随机运气主宰，《随机漫步的傻瓜》值得多看看。准备了一波之后，7月底到8月初面了京东广告、百度自然语言处理部和腾讯AMS，三个都是走完了面试流程，但结果都十分玄学。京东三面很顺利走完而后8月中旬面完HR面确定offer；百度NLP部投的是base深圳，却被北京的部门面，三面经理面聊得很顺畅，然后被玄学挂，复盘猜测是说了优先考虑深圳，后面又被NLP部深圳约面试又是很懵的面试很懵的被挂；腾讯AMS也是一波三折，6月底找内推走的技术大咖(就是这么头铁，渣硕走技术大咖)，7月中开始面试，因为特别重视，每次约面试都把时间后移做足准备，中间熬夜刷了前两百的LeetCode，可以说是研究生最勤奋的一段时间了：一面比较波折，两道算法题一道middle 一道hard，磕磕碰碰做完；二面比较顺利，聊得项目以及一些场景思考题，比较灵活；三面就比较玄学了，面试官比较忙约了三次时间终于面上，面完感觉一般心里就当做凉了。面完腾讯三面，整个人是十分down的，整个秋招提前批可以说几乎无收获，三家公司只有京东过但base北京，收拾心情开始准备秋招正式批。
秋招正式批投了阿里推荐搜索部、美团广告，一如既往的头铁碰大厂业务核心部门。转正申请提交截止日当天，询问得知AMS面试通过等待流程，反复确认后遂放弃申请，而后是漫长的流程，一周后HR面试，中秋后发意向书，国庆后接到正式offer，秋招结束。这期间的一个月里，进入一种奇怪的状态，高度紧张后突然放松难再提起精神准备，腾讯没有正式确认又放不下心，这段期间面试也是波折不断：vivo白菜劝退；平安测评被挂(至今最牛逼的一种挂法)；阿里一面面试官表示通过等二面，过两天看显示已回绝，而后让阿里钉钉捞简历，又简历关挂；华为就是罗生门了，A计划未报名，顶尖计划走的晚面完被转入优招，优招错过笔试不能参加等正式秋招，正式秋招前告知算法HC已无转软件开发，拒绝后联系其他部门，神奇的到深圳参加秋招，三面面试官表示优秀申请15级，至今在备胎池泡着；美团慢慢的面了三面技术加一面HR，拿下offer，但是base又是上海。
总结看来，整个秋招真的是惊险上岸，拿下的offer只有鹅厂在深圳，之前的拖延以及多线并行秋招的节奏十分的混乱，所幸好事多磨，拿到心仪offer。
面经
网上的面经十分多，面向NLP岗的面试经验，推荐别求面经了！小夕手把手教你如何斩下和选择NLP算法岗offer！，当时也是对着准备了一遍。关于面经，我更想说说一些软性的建议，可以参考看看：
(1)节奏
整个秋招找工作，所有人时间总额是一样的，受到的牵制类似(实验室、实习工作)，如何分配时间，避免多线并行的疲惫，是需要好好规划，控制好自我的节奏；自己举例，没有抓住秋招提前批的时间好好刷题和准备知识，导致后面面试还是比较被动的，虽然对于基础比较扎实的同学这些可能不是问题；另外的就是，自主准备还是实习，这也要把握好，今年HC收紧，多数公司实习转正变困难，如果一直实习最后未转正再秋招又很被动了。
(2)信息
找工作，其实是信息不对称的，哪些公司提前批开始和内推渠道，哪些部门缺人，各个公司的部门组成与差异，你面试的部门做的是什么内容。更多的信息就是优势，如何去获取这些信息。公司提前批开始信息和内推渠道，牛客网上可以看；各个公司部门组成与差异，可以去脉脉职言区看看，但注意不要被脉脉的焦虑氛围感染；面试部门是做什么内容，有HR联系就多找HR问问，然后对着招聘JD解读，是可以看出到时会面试哪些东西。秋招的过程中，你的微信好友是肯定会增多一批，每多一个注意沟通也能获取到比较多的信息。
(3)沟通
沟通包括两个方面：一个是面试中和面试官的沟通，面试官考察你问题时，先沟通清楚面试官意图，而后作答，即使不会也要持续沟通和思考，争取面试官引导你，另外沟通时要显得乐于沟通，就不要随便和面试官上头然后开怼了；另一个方面是你的展示，你简历展示的东西，你自我介绍展示的东西，简历的项目介绍需要好好写一下，主要是逻辑有条理，自我介绍上也是注意这一点，自我介绍的另一个点是声音洪亮清晰有自信，如果你说话声音小又显得不自信，无形也会拉低面试官对你项目的印象。
算法工程师的思考
今年算法岗号称诸神黄昏，竞争是十分的激烈，自己最后拿到满意offer运气与努力各占几分也不好判定。很多同学开始思考是否适合走算法路，如果不走算法又该转什么方向，这个可以看看老宋的2021 校招算法岗， 劝退还是继续，讲的比较全面符合今年情况了。
算法工程师的一个重要能力是工程能力，如何定义这个工程能力并提高知乎上也是有许多讨论，这个回答是个人感觉中立 不传播焦虑也不宣传优越感，一条ML-->System-->Web的路线。算法岗诸神黄昏，算法初级职位内卷，如何选择适合自己的方向?
结语：凡是过往，皆为序章
秋招是学生生涯的结束，职业生涯的开始，前路依然长，不焦虑也不能松懈
本文由作者授权AINLP发布于公众号平台，欢迎投稿，AI、NLP均可。原文链接，点击"阅读原文"直达：
https://zhuanlan.zhihu.com/p/105098106
关于AINLP
AINLP 是一个有趣有AI的自然语言处理社区，专注于 AI、NLP、机器学习、深度学习、推荐算法等相关技术的分享，主题包括文本摘要、智能问答、聊天机器人、机器翻译、自动生成、知识图谱、预训练模型、推荐系统、计算广告、招聘信息、求职经验分享等，欢迎关注！加技术交流群请添加AINLP君微信(id：AINLP2)，备注工作/研究方向+加群目的。
