Implementing the LHC on a Whiteboard

A survival guide for passing coding interviews.

Transcript

1. Art and photos by Summer Gray
   Implementing the
   LHC on a Whiteboard
   How to survive
   coding interviews
   of all shapes and sizes
   

   View Slide

2. Why Coding Interviews?
   

   View Slide

3. View Slide

4. This Mismatch…
   ❖ Is keeping programmers from getting jobs
   ❖ Is likely due to a lack of interview training (especially
   compared to the other training we do in our industry)
   ❖ Is fixable
   

   View Slide

5. What Is a Coding Interview?
   ❖ Take homes
   ❖ With or without time limits
   ❖ Technical interviews
   ❖ “Pairing”
   ❖ The dreaded whiteboard
   ❖ Auditions
   

   View Slide

6. What Do I Know About This?
   ❖ I work for NoRedInk
   ❖ We get a lot of applicants
   ❖ Roughly 75% of the candidates fail our “take home”
   ❖ Around 30% of the rest fail their first technical interview
   ❖ I see a lot of what doesn’t work
   

   View Slide

7. We’re hiring
   It’s In My Interest To
   Improve How You Do
   The better candidates do,
   the faster we hire and
   the less I interview
   

   View Slide

8. What This Talk Is Not?
   ❖ Guaranteed
   ❖ A moral judgement
   ❖ All-knowing
   ❖ Respectful of your time
   

   View Slide

9. How Can You Prepare?
   

   View Slide

10. Study
    ❖ Interviews commonly focus on a certain subset of
    programming knowledge
    ❖ This probably isn’t knowledge that you use daily
    ❖ Ensuring it’s fresh in your brain can help
    

    View Slide

11. Concepts Data Structures Algorithms
    Big O Notation Hash Tables Binary Search
    Dynamic Programming/
    Memoization
    Trees/Tries/Graphs
    Bread-First Search/
    Depth-First Search
    Recursion Linked Lists
    Merge Sort/
    Quicksort
    Bit Manipulation Heaps
    Stacks and Queues
    

    View Slide

12. Concepts Data Structures Algorithms
    Big O Notation Hash Tables Binary Search
    Dynamic Programming/
    Memoization
    Trees/Tries/Graphs
    Bread-First Search/
    Depth-First Search
    Recursion Linked Lists
    Merge Sort/
    Quicksort
    Bit Manipulation Heaps
    Stacks and Queues
    

    View Slide

13. http://bigocheatsheet.com/
    

    View Slide

14. Big O Notation
    ❖ A metric used to describe an upper bound on algorithm
    efficiency
    ❖ Can be used for time or space
    ❖ Drops constants and non-dominant terms
    ❖ Adds independent work; multiplies dependent work
    ❖ Not just an academic curiosity! It can help you solve
    problems faster and easier!
    

    View Slide

15. def sum_and_product(nums)
    { sum: nums.inject(0, :+),
    product: nums.inject(1, :*) }
    end
    p sum_and_product(1..10)
    

    View Slide

16. def sum_and_product(nums)
    { sum: nums.inject(0, :+),
    product: nums.inject(1, :*) }
    end
    p sum_and_product(1..10)
    O(n)
    

    View Slide

17. def sum_and_product(nums)
    { sum: nums.inject(0, :+),
    product: nums.inject(1, :*) }
    end
    p sum_and_product(1..10)
    O(n)
    

    View Slide

18. module Fib
    module_function
    def recursive(i)
    return i if i < 2
    recursive(i - 2) + recursive(i - 1)
    end
    end
    p Fib.recursive(10)
    

    View Slide

19. module Fib
    module_function
    def recursive(i)
    return i if i < 2
    recursive(i - 2) + recursive(i - 1)
    end
    end
    p Fib.recursive(10)
    O(2n)
    

    View Slide

20. module Fib
    module_function
    def recursive(i)
    return i if i < 2
    recursive(i - 2) + recursive(i - 1)
    end
    end
    p Fib.recursive(10)
    O(branchesdepth)
    O(2n)
    

    View Slide

21. def shared_count(a, b)
    sorted = b.sort
    a.count { |i|
    sorted.bsearch { |j| j >= i } == i
    }
    end
    p shared_count(
    [35, 55, 13, 49, 40, 59, 27],
    [39, 40, 60, 17, 55, 58, 35]
    )
    

    View Slide

22. def shared_count(a, b)
    sorted = b.sort
    a.count { |i|
    sorted.bsearch { |j| j >= i } == i
    }
    end
    p shared_count(
    [35, 55, 13, 49, 40, 59, 27],
    [39, 40, 60, 17, 55, 58, 35]
    )
    O(b log b + a log b)
    

    View Slide

23. def shared_count(a, b)
    sorted = b.sort
    a.count { |i|
    sorted.bsearch { |j| j >= i } == i
    }
    end
    p shared_count(
    [35, 55, 13, 49, 40, 59, 27],
    [39, 40, 60, 17, 55, 58, 35]
    )
    O(b log b + a log b)
    

    View Slide

24. def shared_count(a, b)
    sorted = b.sort
    a.count { |i|
    sorted.bsearch { |j| j >= i } == i
    }
    end
    p shared_count(
    [35, 55, 13, 49, 40, 59, 27],
    [39, 40, 60, 17, 55, 58, 35]
    )
    O(b log b + a log b)
    

    View Slide

25. A: 13, 27, 35, 40, 49, 55, 59
    B: 17, 35, 39, 40, 55, 58, 60
    Given two sorted arrays, find the elements
    in common. The arrays are the same length
    and each has all distinct elements.
    

    View Slide

26. A Brute Force Solution
    a = [13, 27, 35, 40, 49, 55, 59]
    b = [17, 35, 39, 40, 55, 58, 60]
    shared = a.select { |n| b.include?(n) }
    p shared
    

    View Slide

27. A Brute Force Solution
    a = [13, 27, 35, 40, 49, 55, 59]
    b = [17, 35, 39, 40, 55, 58, 60]
    shared = a.select { |n| b.include?(n) }
    p shared
    

    View Slide

28. Bungling Towards Ideal
    a = [13, 27, 35, 40, 49, 55, 59]
    b = [17, 35, 39, 40, 55, 58, 60]
    i = 0
    shared = b.each_with_object([ ]) do |n, array|
    while i < a.size && a[i] < n
    i += 1
    end
    array << n if a[i] == n
    end
    p shared
    

    View Slide

29. Bungling Towards Ideal
    a = [13, 27, 35, 40, 49, 55, 59]
    b = [17, 35, 39, 40, 55, 58, 60]
    i = 0
    shared = b.each_with_object([ ]) do |n, array|
    while i < a.size && a[i] < n
    i += 1
    end
    array << n if a[i] == n
    end
    p shared
    

    View Slide

30. Bungling Towards Ideal
    a = [13, 27, 35, 40, 49, 55, 59]
    b = [17, 35, 39, 40, 55, 58, 60]
    i = 0
    shared = b.each_with_object([ ]) do |n, array|
    while i < a.size && a[i] < n
    i += 1
    end
    array << n if a[i] == n
    end
    p shared
    

    View Slide

31. Straightforward Linear
    a = [13, 27, 35, 40, 49, 55, 59]
    b = [17, 35, 39, 40, 55, 58, 60]
    c = b.each_with_object({ }) do |n, hash|
    hash[n] = true
    end
    shared = a.select { |n| c.include?(n) }
    p shared
    

    View Slide

32. Straightforward Linear
    a = [13, 27, 35, 40, 49, 55, 59]
    b = [17, 35, 39, 40, 55, 58, 60]
    c = b.each_with_object({ }) do |n, hash|
    hash[n] = true
    end
    shared = a.select { |n| c.include?(n) }
    p shared
    

    View Slide

33. Ruby Can Sometimes Help
    ❖ Minitest::Benchmark supports:
    ❖ assert_performance_constant()
    ❖ assert_performance_exponential()
    ❖ assert_performance_linear()
    ❖ assert_performance_logarithmic()
    ❖ assert_performance_power()
    

    View Slide

34. Analyzing Algorithms
    module Fib
    module_function
    def recursive(i)
    return i if i < 2
    recursive(i - 2) + recursive(i - 1)
    end
    def memoized(i)
    cache = Hash.new { |hash, n| hash[n] = n < 2 ? n : hash[n - 2] + hash[n - 1] }
    cache[i]
    end
    def dynamic(i)
    return i if i < 2
    a, b = 0, 1
    (i - 2).times do a, b = b, a + b end
    a + b
    end
    end
    

    View Slide

35. Analyzing Algorithms
    module Fib
    module_function
    def recursive(i)
    return i if i < 2
    recursive(i - 2) + recursive(i - 1)
    end
    def memoized(i)
    cache = Hash.new { |hash, n| hash[n] = n < 2 ? n : hash[n - 2] + hash[n - 1] }
    cache[i]
    end
    def dynamic(i)
    return i if i < 2
    a, b = 0, 1
    (i - 2).times do a, b = b, a + b end
    a + b
    end
    end
    

    View Slide

36. Analyzing Algorithms
    module Fib
    module_function
    def recursive(i)
    return i if i < 2
    recursive(i - 2) + recursive(i - 1)
    end
    def memoized(i)
    cache = Hash.new { |hash, n| hash[n] = n < 2 ? n : hash[n - 2] + hash[n - 1] }
    cache[i]
    end
    def dynamic(i)
    return i if i < 2
    a, b = 0, 1
    (i - 2).times do a, b = b, a + b end
    a + b
    end
    end
    

    View Slide

37. Analyzing Algorithms
    module Fib
    module_function
    def recursive(i)
    return i if i < 2
    recursive(i - 2) + recursive(i - 1)
    end
    def memoized(i)
    cache = Hash.new { |hash, n| hash[n] = n < 2 ? n : hash[n - 2] + hash[n - 1] }
    cache[i]
    end
    def dynamic(i)
    return i if i < 2
    a, b = 0, 1
    (i - 2).times do a, b = b, a + b end
    a + b
    end
    end
    

    View Slide

38. Almost Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibRecursiveBench < Minitest::Benchmark
    def self.bench_range
    [1, 3, 30]
    end
    def bench_recursive_fib
    assert_performance_exponential do |n|
    Fib.recursive(n)
    end
    end
    end
    

    View Slide

39. Almost Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibRecursiveBench < Minitest::Benchmark
    def self.bench_range
    [1, 3, 30]
    end
    def bench_recursive_fib
    assert_performance_exponential do |n|
    Fib.recursive(n)
    end
    end
    end
    

    View Slide

40. Almost Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibRecursiveBench < Minitest::Benchmark
    def self.bench_range
    [1, 3, 30]
    end
    def bench_recursive_fib
    assert_performance_exponential do |n|
    Fib.recursive(n)
    end
    end
    end
    

    View Slide

41. More Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibMemoizedBench < Minitest::Benchmark
    def self.bench_range
    bench_exp(1, 1_000)
    end
    def bench_memoized_fib
    assert_performance_linear do |n|
    Fib.memoized(n)
    end
    end
    end
    

    View Slide

42. More Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibMemoizedBench < Minitest::Benchmark
    def self.bench_range
    bench_exp(1, 1_000)
    end
    def bench_memoized_fib
    assert_performance_linear do |n|
    Fib.memoized(n)
    end
    end
    end
    

    View Slide

43. More Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibMemoizedBench < Minitest::Benchmark
    def self.bench_range
    bench_exp(1, 1_000)
    end
    def bench_memoized_fib
    assert_performance_linear do |n|
    Fib.memoized(n)
    end
    end
    end
    

    View Slide

44. The Most Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibDynamicBench < Minitest::Benchmark
    def bench_dynamic_fib
    assert_performance_linear do |n|
    Fib.dynamic(n)
    end
    end
    end
    

    View Slide

45. The Most Helpful
    require "minitest/autorun"
    require "minitest/benchmark"
    require_relative "fib"
    class FibDynamicBench < Minitest::Benchmark
    def bench_dynamic_fib
    assert_performance_linear do |n|
    Fib.dynamic(n)
    end
    end
    end
    

    View Slide

46. The classic “Nine Balls” interview brainteaser
    You have nine balls. Eight are the same weight and
    one is heavier. You are given a balance which tells
    you only wether the left side or right side is heavier.
    Find the heavy ball in just two uses of the scale.
    

    View Slide

47. View Slide

48. View Slide

49. Practice
    ❖ Work code challenges
    ❖ http://exercism.io/
    ❖ http://rubyquiz.com/
    ❖ You need to be able to consistently solve Exercism size
    problems in one hour
    

    View Slide

50. More Study and Practice
    

    View Slide

51. Have Some Open Source
    ❖ Be able to point to a published gem
    ❖ It barely matters what it does
    ❖ The most important details are a strong README, clean
    code, tests, and documentation
    ❖ A small gem is fine
    

    View Slide

52. What To Do While Coding?
    

    View Slide

53. Goals
    ❖ Not the same as normal on the job programming
    ❖ You need to avoid being seen as a bad hire
    ❖ If you can avoid major mistakes, you’re ahead
    

    View Slide

54. Tip #1
    Read Carefully
    And/or ask as many clarifying questions
    as needed
    

    View Slide

55. Tip #2
    Follow Directions
    Many candidates do not do this
    

    View Slide

56. Tip #3
    Stop and Think
    You don’t have time not to do this
    

    View Slide

57. View Slide

58. Tip #4
    Think Out Loud
    Magic words:
    • “My first thought is…”
    • “I’m considering the tradeoffs…”
    • “I might circle back to this…”
    • “I don’t know.”
    

    View Slide

59. Tip #5
    Don’t Make
    Offensive Comments
    Show that you have personality and
    a little bit of humor,
    but keep it appropriate
    

    View Slide

60. Tip #6
    Start at the
    Hard Part
    • Don’t begin by designing objects
    • Jump straight to the interesting bit
    • Fake input as needed
    • Ignore UI completely
    

    View Slide

61. Tip #7
    Test the Core
    Pretty much no one tests anything,
    so just a little, at the key place, is a win
    

    View Slide

62. Tip #8
    Avoid Dependencies
    And don’t recreate them manually
    

    View Slide

63. Tip #9
    Write Less Code
    Don’t “gold plate”
    

    View Slide

64. Tip #10
    Ignore “Bonus”
    Challenges
    Seriously, it’s just not worth it
    

    View Slide

65. Tip #11
    Consider Time
    Limits Soft
    Submit on time,
    but resubmit later if needed
    

    View Slide

66. Tip #12
    Show Off Your
    Text Editing Skills
    • Know five editing shortcuts
    • Make one multiple cursors or macros
    

    View Slide

67. Maximizing for Success
    ❖ You probably won’t nail all of my tips every time
    ❖ That’s fine
    ❖ Interviews are hard for everyone and everyone makes
    mistakes
    ❖ Hit as many of the points we’ve discussed as you can
    ❖ Those you get will help
    

    View Slide

68. Thanks and
    I hope you ace your interviews!
    

    View Slide