Various domestic factories
My 21st slag master, I can say that I have interviewed most of the national Internet. I promised group friends to post a comment. Non-Internet and small factories will not post it. Some interviews will be posted to the blog. So I will be more detailed. There may be omissions in the complete record, please refer to it as appropriate.
The answers and experience will be compiled into documents for everyone, but I don't recommend you to read them directly. You can read the previous real questions first, and after reading it yourself, do it in your mind to see what level you can achieve! full version of the interview questions + experience + video have been compiled for everyone, friends in need pay attention to the official account: the future is bright!
Tencent (backend)
One side (30 minutes)
- TCP three-way handshake, four waved hands and state transfer process
- The difference between Http and Https is mainly the process of S
- Algorithmic problem: find the largest prime factor
- Redis data structure and specific implementation method
- The realization process of i++ sentence system
- ACID and isolation level
- Algorithm question: Linked list to determine the ring and find the position of the ring
Two sides (40 minutes)
- Chat project
- Redis distributed implementation principle Three distributed and sentinel mode consistency hash
- Implementation of distributed locks
- Redis ziplist conditions
- C language related (it seems to be a string copy function (forgot))
- Large amount of data and small memory to find the median
- Algorithmic question: handwriting quick sort
Baidu (data research and development)
One side (50 minutes)
- Chat project
- Introduction to spark commonly used operators
- spark shuffle process
- Data tilt and solution
- map join
- Broadcast variable
- Puzzle: 3 people, 3 hats, red, blue, red, red, blue, and red
- The first time everyone didn't know, the second time someone knew
- Certain distribution red, red and blue
- sql: ask for 7th retention
- Algorithm question: Find the target value of an ordered matrix (leetcode 240 questions)
- The difference between XGB and GDBT
- What improvements have XGB made
- XGB's treatment of missing values
Meituan (Big Data)
One side (70 minutes)
- Chat project
- Each sorting algorithm and complexity (let list 10)
- The difference between Http and Https
- The difference between Spark and Hadoop
- Chat project
- Loop control statements for each language (c go python java)
- Storage medium from slow to fast
- Rhetorical question
Two sides (50 minutes)
- Why there is no internship, views on data development
- Why didn't the autumn move have an offer?
- Introduce the hashmap, why the array and the linked list should be converted to a red-black tree, and why should it be the n-th power of 2
- Still hashmap, why can't equal directly hashcode first
- Why rewrite equal
- The difference between b+ tree and b tree
- ACID
- How does Mysql realize repeatable reading
- Clustered index and non-clustered microcosm
- What statement can use the index, how to optimize the index
- Ask the project, why is it so structured
- Spark shuffle process
- Problem: cross reconstruction of ordered linked list, constant space.
- Rhetorical question
Byte (big data)
One side (50 minutes)
- Do the problem: find the peak of the array
- Optimized for binary search
- Problem: Binary addition
- Make it to optimize (then I didn't think of it, it should be an exclusive OR and AND operation, and I have the original question)
- Chat project
- 2 difficult SQL (not made)
- Principle of XGB
- The principle of Kmeans
- The difference between Kmeans and KNN
- The advantages and disadvantages of unsupervised and supervised learning
- Rhetorical question
Didi (Big Data)
One side (50 minutes)
- Chat project
- TCP and UDP three-way handshake TCP state transfer
- The difference between HTTP and HTTPS and the process with more S
- The association and difference between Tcp and http
- The difference between Kafka and RocketMQ
- Spark rdd introduction, shuffle process introduction, blood relationship depends on the difference between rdd and dataset
- Various data structures and implementation methods of redis
- Redis distributed implementation method
- Question: Forgot, it should be a middle question
- Rhetorical question
Two sides (60 minutes)
- Chat project
- Large amount of data (repetition exists), small memory, count the number of repetitions
- Talk about ideas and optimize under the interviewer's prompts
- Talk about kafka allocation principle, partition storage is related to broke
- Talk about the hash principle of distributed related keys, consistent hash, etc. (some forgot)
- Problem: add big numbers
- Situational questions:
- How to use DBSACN to determine the driver's division area
- Didi business introduction.
- Rhetorical question
Three sides (35 minutes)
- Views on data development
- Sorting algorithms
- How to allocate program memory space
- http and https
- Scenario: Combine tracking numbers and use algorithms to achieve and then optimize
- Rhetorical question
Four sides (adding noodles for 40 minutes)
- Talk about the project and ask in detail about the use of each technology and why
- Ask the ability to withstand stress
- Small talk
Xiaomi (algorithm platform)
One side (15 minutes)
- Do the problem: the left view of the binary tree
- Then ask how to optimize the written algorithm.
- Chat project
- Talk about which algorithms are commonly used (then it ends directly)
Two sides (40 minutes)
- Chat project
- Question: Reverse the linked list
- The difference between XGB and GDBT
- ROC, AUC, accuracy rate, recall rate calculation formula
- Tree model tuning method
Good future (big data)
One side (50 minutes)
- Chat project
- hive: internal table, external table, data skew and resolution, lock table
- spark: The difference between reducebykey and groupbykey
- flink: How to ensure consistency
- Speaking of check and barrier
- sql index and difference
- Speaking of gathering and non-gathering
- Innodb and myism briefly talk about
- Do the question: the consumption from the top left to the bottom right is minimal
- sql: Seeking people who log in consecutively for seven days
- Rhetorical question: business content and evaluation
Two sides
- How does the operating system work?
- Then ask how to manage memory and file management, process management
- Talk about how to implement Hadoop distributed
- Talk about master-slave replication and election (some knowledge in Google’s three major papers)
- Talk about how RDD is distributed
- Do the problem: the second largest number of the array
- How Kafka realizes distributed
- Talking about broke and partiton, I feel that the interviewer is not satisfied
- Let handwritten network programming realize service scheduling of various data structures of redis
- Rhetorically asking which directions need to be improved
Ctrip (data development)
One side (30 minutes)
- Chat project
- Ask the algorithm, logistic regression, and the calculation process of svm
- The principle of feature importance calculation of xgb
- Other algorithmic problems of the project
- Database indexes: clustered index and non-clustered index
- Question: Find the intersection of two arrays
SF Express (Big Data)
One side (40 minutes)
- Java polymorphism concepts, abstract interfaces and abstract classes
- java hashmap and underlying principles
- Which of java sleep and wait will release resources
- Do the problem: Binary tree left view
- An unordered heap becomes a big root heap. Time complexity
- Spark RDD and dataframe, data skew and solutions.
- Spark memory model
Can reducing the number of parallel tasks reduce memory pressure?
My interview experience sharing (super detailed)
First of all, the basic piece must not be left. The so-called high-rise building is based on this principle, including mybatis, springboot, springmvc, and even springcloud, which need to be basically mastered. My words are to choose the video method to study and review. If you have any If you need this aspect, I can share the video with you!
Data Structures and Algorithms
Then there is algorithmic questions. I think that algorithmic questions still occupy a large part of the interview ratio. I usually read the algorithm by watching the video first, and then brush it by myself after watching it, so that the impression and understanding are better!
Brush eight-legged essay
I believe that I don’t need to say that everyone knows his proportion in the interview. Almost no matter which interview you go to, you will ask, how do you accumulate experience? My answer is to review the question, review it by yourself + watch the video explanation, and hold on for a few months, I believe you will be surprised at your changes!
Brush interview questions
I usually choose to go to Niu Ke to see this a few days before the interview. There are still many big guys who summarize it very thoroughly. Of course, if you are lazy, I have compiled a set of 2021 top Internet companies. The interview questions are all explained in the form of video, which is relatively easy to understand!
Reading book
This is the most primitive and relatively efficient learning method. What books you read, this is based on your own situation, what you need to add, I believe you definitely know better than me! Of course, as mentioned above, if you want to be lazy, I also have a set of 150-episodes of the latest version of the Alibaba Development Manual (Songshan version) video details, whether you are a cute new or a big guy, you can choose to check it out and pick it up if you need it. !
Interview questions summarized by myself
I have interviewed so many companies for so many years, and I usually like to take notes and write blogs. I have accumulated a large number of first-line interview questions here. I wrote them down and marked my understanding and answers. Full version I have sorted it into a PDF file, friends in need pay attention to the public account: a bright future, take it yourself! , including all the above videos and materials, please pick up!
at last
Let’s share a little bit of experience. Whether you are changing jobs, looking for a job, or doing an internship, you can always treat this interview as your last interview. Don’t think that you have a way out here, and you can meet again there. It will only affect your performance. Taking every interview seriously, you are responsible for yourself and the company you are interviewing for. Finally, I wish you who are looking for a job successfully interviewed and get a satisfactory offer, and you who do not need to find a job have a smooth job and a successful career! ! ! Think the article is helpful to you, remember to like it!
**粗体** _斜体_ [链接](http://example.com) `代码` - 列表 > 引用
。你还可以使用@
来通知其他用户。