Training Turk

Customer Success Engineer (India)

$30k–45k/yr · Mercor · Full time

A Customer Success Engineer investigates and resolves technical issues for users of an AI hiring platform, debugging across web and AI systems.

What you would do

  • Receive and reproduce talent-reported bugs and issues end-to-end
  • Debug via telemetry, logs, network monitoring, and database queries to identify root causes
  • Triage issues, escalating engineering problems and resolving others via configuration or user guidance
  • Document solutions and create runbooks to improve resolution speed and reduce repeat issues
  • Surface emerging patterns and product risks to the engineering and product teams

Who they want

  • CS degree from a top-tier school or prior experience at a high-growth tech startup
  • 2-5 years building or supporting modern web applications like React, Node, Flask, or similar
  • Comfortable with AI systems: hands-on experience with LLMs, agents, or generative models
  • Understanding of AI model outputs, failure modes, and how modern AI APIs work
  • Native English proficiency and ability to communicate with precision while maintaining an upbeat tone

Main skills

Saas debuggingWeb application troubleshootingLLM and generative AI

What the interview asks about

  1. 1.Debugging across SaaS and AI stacks

    A candidate must be able to navigate logs, telemetry, and database queries to find the root cause of user problems in production, especially when AI components are involved.

    For example: “A user failed an interview after the AI agent misunderstood their answer. How would you debug this? What logs would you check to distinguish between a hallucination and a prompt engineering issue?”

  2. 2.Triage and sound judgment

    The role requires deciding what is a true product bug versus what can be handled through configuration, documentation, or user clarification - this judgment directly affects team velocity and user satisfaction.

    For example: “Three users can't submit answers after 2 PM PT on a specific payment tier. Is this an engineering defect, a system limit, or user confusion? How would you determine which?”

  3. 3.AI model behavior diagnosis

    Since the platform uses AI for hiring decisions, you must understand failure modes and edge cases in generative models - knowing when a model is hallucinating versus when it's a data pipeline problem is critical.

    For example: “An AI interviewer gave a candidate an impossibly difficult follow-up question that seemed unrelated to their previous answer. What would you investigate to determine if this was a prompt chain error, a fine-tuning problem, or a legitimate model output?”

  4. 4.Communication and documentation

    Your support conversations are often users' first impression of the company, and clear runbooks prevent the engineering team from fielding the same questions repeatedly.

    For example: “You resolved an issue by adjusting a LangChain configuration. Now write me a one-paragraph explanation suitable for both a non-technical user and for your runbook, explaining what happened and why we changed it.”

  5. 5.Production investigation with telemetry

    Reproducing bugs in live systems requires methodically gathering data from logs, network requests, and database snapshots, especially when timing or system state matters.

    For example: “A user from India reports their interview took 45 minutes to load. You know the issue only occurs when overlap with Pacific Time hours is minimal. What telemetry would you pull, and how would you reproduce and diagnose this latency?”

A task you may get

A user's interview failed after their connection dropped mid-response. You have logs, submitted answers, and network data. Is this a connection bug, model timeout, or user guidance issue? Explain your diagnosis.

How to prepare

  • Practice debugging a React or Node application using browser dev tools and server logs
  • Try interacting with at least one LLM API (OpenAI, Anthropic, or similar) and experiment with failures like token limits or malformed prompts
  • Review how modern SaaS platforms handle edge cases like network timeouts, partial data submission, and race conditions
  • Write a short troubleshooting guide for a technical issue you recently solved, using both technical and user-friendly language

The facts

Pay
$30k–45k/yr
Hours
Full time
Where
Remote
Field
Software Engineering
Posted
3/12/2026
Places left
2

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