Autumn recruit summary
- The autumn recruitment is over. From June to July to November, from the difficult preparation stage to the busy interview, I have a lot of emotion. Basically, I am dealing with companies in Hangzhou. The offers currently available include NetEase Entertainment, Thunder, Huawei, Haikang, Pinduoduo, Didi, 51, and most of them are data research and development positions. Write a summary of experience and give back to everyone.
In addition, I am here to prepare interview materials for first-line manufacturers and my original super hard-core PDF technical documents, as well as a number of interview questions from large manufacturers that I have carefully prepared for you (constantly updated). Friends in need click here to remark csdn , Just download it yourself! Hope everyone can find the job they want!
Job selection
- Before preparing for the review, you should first determine your own job direction, and then look for the interviews and materials review. At that time, I chose two directions: data mining algorithm and data research and development, because different companies have some differences in the specific requirements of the position, which caused me to be a little confused. I have been in contact with engineering and algorithm competitions myself, but I still have relatively little understanding of algorithm principles, so some backed down and chose the direction of data research and development. In fact, the salary of machine learning is more attractive this year. Judging from my experience, I think I can find a good job. Of course, this is something to do. After determining the position, it is how to prepare.
Review preparation
-
I probably started preparing in June. The biggest difficulty at the beginning was that I didn’t know how to prepare for good results. I saw the students from other laboratories start to prepare after the year, and the autumn recruitment began around July in advance. I was very anxious. In fact, if I want to come now, it is the key to steadily review. Dachang has relatively higher requirements on the foundation, followed by its own bright spots.
-
The first level of the interview is usually to tear the code by hand. In fact, the code question can be used to deal with most interview algorithms by pointing the sword twice to the Offer, because it is impossible for you to write a long code during the interview. Learn enough to brush leetcode.
-
For other knowledge points, you can save all the knowledge of the corresponding positions first, and then summarize the important knowledge points. Knowledge point formation system is more important, and you can draw a system diagram and review it continuously. For example, the JVM part, collection class part, and concurrent part of Java are separated to form a knowledge system. Prepare all the questions in the interviews you have saved. In fact, the interview is not false.
-
Others
The experience of summer internship is very important, but it is a pity that our laboratory does not offer internships.
The opportunity to approve in advance should be grasped well, that is, one more opportunity. In fact, Dachang may still make up for it in November.
Prepare early and show your highlights to the interviewer. -
Face Sutras
Share a few typical face sutras of your own face.
Five aspects of Ant Financial data research and development
- Ant is my first face, and the most regrettable one. I regret that I was not ready at that time. I probably met five sides, but because the review was not good enough and the interview performance was not good enough, I didn't get the offer. Ali is all online, and it takes a long time before and after.
one side
- Self introduction
- project
- Design the recommendation system, algorithm and architecture implementation of Alipay APP sliding bar
- How to obtain the characteristics of user trajectory, click item 7-5-2-4, how does word2vec construct sentence characteristics
- GC
- Linux 10 processes and 10 threads, how to schedule the system
- Thread pool, poolsize 5, what happens when the sixth thread comes in
- Java lock mechanism, pessimistic lock, optimistic lock
- Data structure problem, frog jumps the stairs, binary tree sequence traversal
Two sides
- Tianchi competition items and ideas, why choose XGBoost
- What is the recommendation system? Tell people who don’t understand it at all.
- The shortcomings of collaborative filtering, how to do it now popular (collaborative filtering is currently more used for recall)
- Talk about logistic regression, principle and derivation
- Synchronized. Implementation of the lock at the operating system level
- Concurrent packages in Java
- Linux process scheduling
- Network related protocols
- Zookeeper *** algorithm, Paxos
- How is the source code level to store a file on HDFS with the command line?
- Basic Linux commands, how to check memory usage, etc.
- The difference between dynamic programming and divide and conquer
Three sides
- Play the project by itself, ask for details
- What kind of person you think you are, what other people think of yourself
- Have you had any major setbacks?
- I have done the most fulfilling thing
All sides
- Talk about a project
- Detailed questions in the project (how to do online A/BTest, how to optimize the results)
- Highlights of the project
- There are numbers in a circular linked list, how to find the specified number as quickly as possible
- A file containing English novels with the two most frequent words
- The difference between notify() and sleep() (it should be wrong, wait() and sleep()), where did you see it in the source code (blocking queue)
- Thread pool, corePoolsize 5, max 200, when there are 10 threads in the thread pool
- The blocking queue is full and the threads are over 200, what will happen
- How the thread pool is actually used
- Do you have anything to ask me
- Why didn’t you go for an internship? Are you unable to go out from now until graduation
Five sides
- Self introduction
- What does the Github library do
- What did each project do
- Mapreduce Join
- Mapreduce shuffle
- Remote code question, input one n, find the number of 1 in each digit of 1~n
NetEase Hangzhou Research Big Data Development
one side
- Self introduction
- Mapreduce principle
- What to do if the data is skewed
- For example, how to customize the partition and code of the statistical data in Guangdong Province? (I think I made it quite clear, and the code must be written)
- Recommend system project details, roles, and what to do
- How to adjust the Huawei Software Elite Challenge
- Code, the intersection of two sets, how is the fastest
- Reverse output of string, implemented by recursion
- Know what ETL tools
- dataX architecture
Two sides
- Self introduction
- Project details
- Mapreduce principle
- How to determine splitsize, algorithm
- What to do if the data is skewed
- What is the process without shuffle
- What big data components know
- The difference between data warehouse and database
- Do you understand the principle of database
- Do you know the database paradigm
- sql question, sort by a column, and increase the sort number (saying how to do mysql, hive a bit forgot row_number did not explain, it is a pity)
- sql question, one column is numbered sequentially from 1 to n, one column is various categories, output a category number is more than four consecutive data rows
- Code question, the intersection of two sets..., what are the methods
- What hobbies do you usually have
NetEase Thunder and Fire Data Mining
one side
- The difference between a += 4 and a = a+4, a is byte type
- Integer data comparison
- Python merges the two lists to reorder (Python has not been typed for a long time, and the basic syntax has been forgotten a lot) and I said how to do it.
- Linux two different text statistics, delete directory files, delete duplicate lines
- Code java string reverse
Two sides
- Recommended items
- A SQL table is compressed in gz format, and the query will have several maps and several reduce
- What compression format solves the gz non-fragmentation problem
- The cluster mr task is assigned to a node and has been timeout. (Network, firewall issues)
- Github
Three sides
- Recommended items
- Code outputs all cases where the sum of consecutive numbers is equal to a certain value
- Known algorithm
- LR principle, why use maximum likelihood
- Random Forest Principle
- Bagging principle
- Difference between GBDT and XGBOOST
- Why choose XGBOOST for the competition
- Boosting
All sides
-
Introduce yourself, let alone technology
-
What benefits will I bring to Thunderfire
-
The most unacceptable state will consider leaving
-
Recommendation system application
-
Personal strengths, setbacks, etc.
to sum up
Finally, I prepared the Java architecture learning materials for everyone. The learning technology content includes: Spring, Dubbo, MyBatis, RPC, source code analysis, high concurrency, high performance, distributed, performance optimization, microservice advanced architecture development, etc. Friends in need can click here to remark csdn and download by themselves! , Another reminder, full review is the key to eliminating your nervous state of mind, but if you review enough, you will naturally have more confidence in the interview process.
There are also Java core knowledge points + a full set of architect learning materials and videos + first-line interview collections + interview resume templates + Alibaba Meituan Netease Tencent Xiaomi Iqiyi Kuaishou Bilibili interview questions + Spring source code collection + Java architecture actual combat electronics Share the book with everyone for free!