The repo includes duke course materials and my peronsal suugestions, and track what I learned and achieved since joined Duke.
conatct: zw15724584198@gmail.com

• A trip to Florida with friends majoring in Economy and Jack. We had a detailed plan ahead of the trip visiting Orlando, Universal Studio, Key West, Wetland, seeing alligator
• Another trip with CS friends to Zion national park. We rented RV and had a greate experience. It was a blast
• Start to do some real stuff.

• no classes since i satisfied all the requirments and choose to graduate with project/thesis.
• Auditing danyang's seminar class: Data Center systems(https://courses.cs.duke.edu/spring22/compsci590.4/). Read many classic paper which onctains three categories.
  • data center networking
  • machine leanring system
  • system
• working on the project
  • extend work on Lightning(https://github.com/wz30/lightning): distributed lightning: master/worker model with hashing slots.
• UCLA with Yuval Tamir(https://web.cs.ucla.edu/~tamir/)
• TA in programing interview class (introduction to leetcode and some basic data structure)
• EAD is approved by filing online. my.uscis.gov

• socket fuzzing reserach

• CS 514 computer network
  • Midterm exam is harder than final exam. The exam is testing if you really understand the concept, not mechanically remember the concept. Quite different from Tyler's computer security
  • My project: analysis on various TCP congestion algorithm.
  • Homeworks are useful materials to study the key points of the lectures
  • taught by Bruce Maggs and Shane Zhang
• CS 550 computer architecture
  • taught by Daniel Sorin
• CS 590 secure software system
  • taught by Mattew Lentz
• Academic presentation by Brad
• TA in 201 Data structure
• Intuit full time return offer

• Summer internship in Intuit(IDLM team)
  • 1st project: graphql Linter project
    • First stage: build a tool to check schema before runtime.
    • Second stage: deploy a container to host such tool and serving the GraphQL Schema Playground webiste. User can gain the best practice from the website.
  • I learned Kubernates by myself through a real example: sentiment analysis.
  • 2nd Project: Migration from IKS1 to IKS2 (Intuit Kubernates Services)
    • this is a devops task
    • tech: jenkins peipeline, ArgoCD, Kubernates
• Research: network hint
  • figure out the better way to manage the traffic flow.
• Learning distributed system
  • Now I am focusing on Raft .
    • passed 2A
• Learning distributed System by Mark from Cambridge.

• edge computing: empower IoT and cloud computing. Edge node are like the middle layer between IoT and cloud server

  • Me and my partner work on the research project: practical data cleaning in Federated Learning.
  • We are required to read many papers and prepare for reading quiz for each paper.

• ECE 568 robust server software

  • 1st project: django to create rider share web app.
  • 2nd project: web cache proxy server (using multithread, lock and socket programming)
    • cache overview: https://www.mnot.net/cache_docs/#ABOUT
    • good tutorial about socket programming: http://beej.us/guide/bgnet/html/#select
      • recv can receive header or header+contents. it depends
    • mocklab to customize the response header and contents: https://app.mocklab.io/mock-apis/l7wwd/stubs/54f6eb4b-e621-44ba-96bc-9903ccbb4b4a
  • 3nd project: stock exchange server software: TA will test the propgram in details. Another requirement is loading testing.
  • final project: mini UPS and mini AMAZON. Google protocol buffer is a hard part. Testing is also hard.
  • Midterm is hard sicne we have short time to complete and there are 7 questions. The exam is testing your basic and deep knowledge on the topics. In addition, how to apply the knowledge is another requiremnt for the exam.
  • Final exam is similar to Midterm exam but with more time. It focues more on security topic. Not hard one.
  • What I learned from this class:
    • how to create a robust server software in various aspects?
    • Class assignment involves lots of codings and takes more time than I thought. Expertise on the programming language/framework is a prerequisite.

• interview class(1.5 credit taught by Owen): leetcode and hackrank problem

• writing class: problem-solution based paper and topic are field related.

• TA CS 201: quite a lot of logistic works.

• Until 4/5/2021, I am not doing very well in this semester and I just figure out why that would happen. The main reason is that I made myself super busy and that hurts my confidence. TA in 201 is not a good choice for me since it takes up lots up time to grade the homework and ECE 568 projects take more time. Anyway, I need to catch up where I left.

• Since school ends Fall early, we have two months winter break ever. It is the longest one ever.
• I had an ambitious and detailed plan but it did not work very well.
  • keep working on the research every week but not really productive since I was stuck on this and did not yield any good outputs. Vicious cycle begins. BTW, the research topic is about container and networking. it is a good and a promising area. However, I did not have enough knowledge on that.
  • OSTEP book: three easy pieces book. Finished around 20 or less chapters. It is a great book to learn and start operating system.
  • first couple chapters of Linux programming intereface.
  • leetcode problem , not a lot.
  • ski once and that was a good experience
  • container and k8s learning.
  • Teaching Charles and Cindy every week.
• Overall, it is not a productive winter break and did not work a lot every day. Thera are couple improvements.
  • workout and shower are the best and feasible way to recover from "Pandenmic mood"

• machine leaning in CS department
• computer security in ECE
• Introduction to AI in CS
• Algorithms in real world undergrad courses in CS
• SLIM - network research

• final exam: Since COVID 19, it is a 3 hout exam. And the problem is quite comprehensive, including all the core parts taught in class.
  • submodular
  • HMM: computer viterbi, forward, backward.
  • Baynes Net:
    • best case and worst case to compute p(XiXj) in terms of time complexity.
    • reduce 3SAT to problem "in baysian network, at most two parents, is p(x)>0"
  • Linear sperable: given specific linear funtion such as w1x1+w2x2+w3x3 + w4 >0 , check if given formula is linear seperable or not. E.g. X1 or (X2 and X3)
  • Using Baynes net as linear classfier and derive the expression to compute the label
  • ML: understand what trend of plot when the factors change?
    • learning rate is too high/low
    • iteration is not enough
    • training set is too small
    • stuck on local optimal