TIL > job-interviewing >

Turbo repo team interview

Days before

  • algo

  • Dropbox pracitce algo

Determine efficiently if a folder has changed.

Time stamp and size diff.

The if those fail do a hash compare as a last resort.

Look at algo of scheduling code. Try to Calculate big O of it and think of algos that can improve it

Review stripe algo. Cache busting algos. File access algos.

Content CDN algos.

Calendar algo

What problems is Turborepo trying to solve?

Turborepo is a high-performance build system for JavaScript and TypeScript codebases. -What are the engineering challenges?

  • Q: What

How can my unique skills & talents help provide value to this team (and make their lives easier?)?

  • devrel work
  • ci & testing
  • ease into issues work and more complex work & troubleshooting.
  • Docs, clarity, demos.
  • Codemods. Ast. Diff tooling.

I could start out as devrel and then ease into more repo work... Be active in the discussions board...

Things to highlight about myself (I'm scrappy)

  • I'm a generalist & solver.
  • I love debugging across systems. I really like getting my hands dirty to understand the pieces better. Eager to learn more parts of the stack.
  • I love being able to take ownership of a problem / domain
  • I'd love some mentorship across any of the following:
    • product / project decisions
    • how to make sound technical decision amid ambiguity
    • staff/principal level engineering decisions. Where and how to draw boundaries across problem & solution spaces well
  • Currently reading books on product & project management to round out my decision making ability better and managing risk
  • Care deeply about good DX (dev experience) and UX
  • I've spent a lot of time with metrics and getting meaningful proxies for things like app health
  • Love teaching & speaking & presenting. I have 3 courses online (React, ESLint, Redux).

Finally

  • I crave growth across some vertical. Whether it's technical, or product, or decision making.
  • I like operating 1-2 steps outside of my experience and comfort area
  • I like having loose direction "solve X" and then being able to bounce off / collaborate on where the problem & solution boundaries are and setting appropriate appetite and scope for the size & duration of the problem & investment.

Questions to answer

  • how does Turbo cache things locally? Or does it just do this for the network? Q: How does it skip running those build tasks again?

Q: does it look at sources? See what tasks affect what files? Q: Attach itself to the process?🇶🇦

Research:

  • Semaphore
  • mutex lock

TODO Build:

  • read a bunch of files. Then perform some operation
    • use a mutex? Would love if we could fork & thread into the processing, but the resulting graph should be shared.

Algo learning:

  • lots of inserts, few reads? Use Linked list

  • lots of read (at address), few inserts? Use array.

  • Simple search O(n)

  • Binary Search O(log n)

Q: Can we do a dry run first and build a graph of necessary operations? BFS?