Training Turk

Competitive Coder

$45–65/hr · micro1

You design and validate competitive programming problems using C++ to help train AI models on algorithmic reasoning.

What you would do

  • Create original problems absent from online judge platforms, ensuring they test distinct algorithmic skills
  • Develop reference solutions and robust checker programs that differentiate correct from incorrect approaches
  • Write comprehensive test suites covering multiple solution strategies, performance constraints, and edge cases
  • Document problem specifications and test logic clearly to guide external review and evaluation
  • Iterate on problem design based on feedback from reviewers and problem setters

Who they want

  • Expert-level competitive programming background with track record of problem-solving across platforms like Codeforces
  • Advanced C++ proficiency for implementing efficient, maintainable checker logic
  • Strong analytical abilities to identify problem constraints and multiple solution paths
  • Meticulous attention to correctness, ensuring every test case is justified and intentional
  • Ability to work independently in remote settings while communicating technical details in writing

Main skills

Basics C++C++Competitive Programming

What the interview asks about

  1. 1.Designing multi-solution problems

    A problem that accepts only one solution pattern severely limits its usefulness for training AI, which needs exposure to diverse algorithmic approaches.

    For example: “You design a graph problem; a breadth-first search solves it optimally, but depth-first search also works if memoized correctly. How do you ensure your checker accepts both?”

  2. 2.Handling floating-point precision

    Many geometry and numerical problems fail because checkers reject correct solutions due to strict equality checks or rounding errors.

    For example: “Your problem asks for the area of a polygon. Contestant A uses trigonometry getting 12.0000001, contestant B uses the shoelace formula getting 11.9999999. How does your checker handle this?”

  3. 3.Identifying problem difficulty boundaries

    An AI model learns nothing from trivial problems, but an impossible one causes training to fail; your job is finding the sweet spot.

    For example: “You set a time limit of 2 seconds for a problem. When you test it, your solution runs in 0.15 seconds. What do you do next to verify the difficulty is appropriate?”

  4. 4.Preventing hardcoding exploits

    AI systems are notorious for pattern-matching rather than reasoning; test cases must make it impossible to pass by remembering output patterns.

    For example: “Your problem involves finding the k-th smallest element. You only test with k=5, k=10. What's the vulnerability, and how do you fix the test suite?”

  5. 5.Communicating design rationale

    Reviewers need to understand not just what the problem is but why it's worth solving, what skills it targets, and whether it's suitable for AI training.

    For example: “A reviewer questions whether your problem tests useful real-world skills or just exercises obscure edge cases. Describe how you defend your problem's pedagogical value.”

A task you may get

Design a coding problem from scratch that tests a specific algorithmic skill (e.g., dynamic programming, graph traversal), then write a C++ checker that accepts multiple valid solution approaches.

How to prepare

  • Study recent Codeforces problem editorials to understand how expert problem-setters articulate solution approaches and test coverage
  • Practice implementing checkers for existing problems, focusing on handling floating-point comparisons, multiple data types, and corner cases
  • Review common pitfalls where test suites fail AI systems: hardcoded patterns, weak edge cases, over-specified solutions

The facts

Pay
$45–65/hr
Open to
Bangladesh, Hong Kong, India, Indonesia, Japan, Kazakhstan, Kyrgyzstan, Malaysia, Pakistan, Philippines, Singapore, Sri Lanka, Taiwan, Thailand, Uzbekistan, Vietnam, Austria, Belarus, Belgium, Denmark, France, Germany, Greece, Italy, Netherlands, Portugal, Russia, Spain, Switzerland, United Kingdom, Argentina, Brazil, Chile, Colombia, Mexico, Peru, Algeria, Bahrain, Egypt, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Palestine, Qatar, Saudi Arabia, Tunisia, United Arab Emirates, United States, Canada, Nigeria, Kenya, South Africa, Ghana, Ethiopia
Field
Software Engineering
Role type
Expert
Posted
9/16/2026
Places left
300

We wrote this page from the public micro1 listing. It may be out of date, so read the full posting before you apply.