Training Turk

Software Engineer, Full Stack — India

$25–30/hr · Mercor · Full time, 40 hours a week

Full-stack software engineer building applications that test frontier AI models using multiple programming languages and cloud infrastructure.

What you would do

  • Build full-stack applications with skills spanning Python, Java, Rust, C++, C#, and TypeScript
  • Integrate pre-release AI model APIs with surrounding tools, evaluation frameworks, and observability
  • Diagnose integration failures and model behavior issues, translating findings into technical reports
  • Prototype quickly against shifting requirements, delivering working code with minimal specification
  • Collaborate on architecture, code reviews, and documentation standards with engineering teams

Who they want

  • 3+ years building and shipping production software at top-tier organizations
  • Production proficiency in at least two of Python, Java, Rust, C#, C++, or TypeScript
  • Full-stack hands-on experience: backend services, APIs, React or modern front-end, databases, cloud
  • Demonstrated ability to onboard onto unfamiliar codebases and contribute working code quickly
  • Strong written communication skills and reliable availability for 40 hours weekdays

Main skills

Python, java, rust, c#, c++Full stack developmentRest apis and backend services

What the interview asks about

  1. 1.Multi-language production experience

    This role intentionally mixes language ecosystems. You need to work confidently across Python, Java, Rust, C#, and C++. Interview checks depth, not just syntax familiarity.

    For example: “Tell me about a production system you built in one language, then later worked on in a second. What surprised you when switching? What idioms took time to learn?”

  2. 2.Rapid codebase onboarding

    You'll join a moving target: unfamiliar codebases, shifting specs. Interview assesses how fast you become productive. Do you read code methodically? Do you find the right people to ask?

    For example: “You've been given a pre-release API integration task in a codebase you've never seen. It's due in three days. What's your first step? How would you unblock yourself if docs were sparse?”

  3. 3.API integration and error handling

    Pre-release model APIs are brittle and will fail. Interview checks whether you design defensively: contract assumptions, timeouts, retries, fallbacks, and diagnostic logging.

    For example: “You're integrating a new model API that can time out or return partial results. How would you structure your integration code to handle both failure modes? How would you log what happened?”

  4. 4.Full-stack architecture decisions

    You choose the front-end framework, database schema, deployment strategy. Interview probes how you think about trade-offs: latency, scalability, maintainability.

    For example: “You need to build a tool that lets researchers run models on batches of data, monitor progress, and export results. Would you use a relational database, document store, or cache layer? Why?”

  5. 5.Translating technical observations

    Research teams rely on your clear communication to fix issues. Interview checks if you can explain what you found, why it matters, and what's needed to resolve it.

    For example: “You notice a model API returns correct answers slowly when input text exceeds 5000 tokens. How would you document this for the research team? What details matter?”

A task you may get

Design a full-stack tool that submits text batches to a pre-release model API, collects results, and lets researchers compare outputs. Sketch the backend service, database schema, and front-end UI. Note failure modes.

How to prepare

  • Review production experience in your two strongest languages. Prepare detailed examples of systems you built, not just features you added.
  • Practice describing architectural trade-offs: why you chose a particular database, framework, or deployment model, and what you'd do differently.
  • Prepare one example of onboarding onto an unfamiliar codebase. How quickly did you become productive? What obstacles did you face?
  • Study API integration patterns: error handling, circuit breakers, retry logic, and how to design contracts defensively.

The facts

Pay
$25–30/hr
Hours
Full time, 40 hours a week
Where
Remote · India
Open to
IND, IND
Field
Software Engineering
Posted
8/10/2026
Places left
10

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