New opportunity

AI Support Engineer

Cognition · Singapore

About this role

Cognition seeks an AI Support Engineer in Singapore to resolve complex technical issues for customers using Devin, its AI software engineer, and Windsurf, its AI-native IDE. The role focuses on investigating ambiguous, environment-specific problems in real-world software development workflows and driving them to resolution while handling a high volume of incoming issues.

Responsibilities

Reproduce, diagnose, and resolve customer issues involving codebases, cloud infrastructure, CI/CD, APIs, containers, version control, IDEs, and enterprise deployments.
Analyze logs and traces, follow code paths, correlate system behavior, isolate failures, and develop root-cause hypotheses. Escalate product or infrastructure issues to engineering with reproduction steps, logs, environment details, and technical findings.
Communicate clearly with customers about investigations, best practices, workarounds, deployment patterns, and product capabilities. Build internal playbooks, tools, automations, and documentation; identify recurring failure modes and share feedback with product, engineering, and deployment teams.
Support release testing and quality efforts across customer environments, programming languages, IDEs, and workflows.

Required qualifications

4+ years in a technical role such as software engineering, infrastructure engineering, solutions engineering, technical support engineering, or developer tooling.
A bachelor’s degree or higher in Computer Science, Software Engineering, or a related technical field, or equivalent practical experience.
Working knowledge of Linux, Docker, Git, CI/CD, networking fundamentals, and cloud platforms such as AWS, GCP, or Azure. Ability to debug incomplete or ambiguous reports, analyze behavior across distributed or multi-component systems, and read code in languages such as Python, TypeScript, Java, or Go.
Strong written communication, sound debugging judgment, and the ability to work urgently across changing technical issues.

Preferred

Experience building support or QA tools and automations; familiarity with LLMs, AI-assisted development tools, and IDEs; high-volume technical support experience; cross-stack debugging; and direct collaboration with engineering on escalations.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

Technical support engineeringDebuggingRoot-cause analysisLinuxDockerGitCI/CDAWSGoogle Cloud PlatformMicrosoft AzureNetworkingAPI troubleshootingLog analysisDistributed systemsPythonTypeScriptJavaGoCustomer communicationTechnical documentationAutomationRelease testingLLMsIDEs

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Reproduce issues across customer environments
The role investigates Devin and Windsurf issues involving codebases, cloud infrastructure, CI/CD, APIs, containers, version control, IDEs, and enterprise deployments. Be ready to explain how you reconstruct an environment-specific failure from an incomplete report.
Establish evidence-based root causes
The posting emphasizes reading logs and traces, following code paths, correlating system behavior, and testing hypotheses to isolate failures across multiple components.
Maintain quality at high issue volume
Cognition calls for resolving a steady stream of technical issues urgently without losing investigation quality or clear customer communication, including explanations of workarounds and product capabilities.
Make engineering escalations actionable
Product or infrastructure escalations should include reproduction steps, logs, environment details, technical findings, and root-cause hypotheses so engineering can continue the investigation.
Turn individual cases into broader improvements
The role builds playbooks, tools, automations, and documentation; shares recurring failure modes with product, engineering, and deployment teams; and supports release testing across customer workflows.

Likely challenges to think through

Ambiguity versus response speed
Likely preparation challenge, not a confirmed employer problem: incomplete reports may demand quick action before the cause is clear. Prepare a triage approach that identifies the most informative missing details, tests a narrow hypothesis, and gives customers a useful update without overstating certainty.
Product defect versus environment failure
Likely preparation challenge: symptoms can cross application code, IDEs, containers, networking, CI/CD, and cloud configuration. Practice tracing one failure across layers and documenting the observations that justify either a customer-side workaround or an engineering escalation.
Casework versus lasting fixes
Likely preparation challenge: a high issue volume may leave limited time for playbooks, automation, and release-quality work. Prepare an example of how recurring cases could be grouped into a reusable diagnostic step or targeted test while preserving time for urgent investigations.

Practice your answers

Technical practice

How would you reproduce an environment-specific issue reported with incomplete details?

What to coverSuggested preparation: Explain which details you would request about the workflow, environment, expected behavior, and observed failure. Describe how you would establish a baseline and change one variable at a time to distinguish a product issue from an environment-specific one.

How would you use logs, traces, and code paths to isolate a failure across multiple components?

What to coverSuggested preparation: Describe how you would correlate events across components, identify where observed behavior first diverges from expectations, and test competing root-cause hypotheses. Distinguish evidence from assumptions when a trace or log is incomplete.

What would you check when a workflow works on a Linux host but fails inside a Docker container?

What to coverSuggested preparation: Compare runtime configuration, dependencies, permissions, filesystem access, and connectivity before attributing the failure to the application. Explain how you would make the difference reproducible.

How would you investigate a CI/CD failure that appears only after a Git change?

What to coverSuggested preparation: Compare the failing run with a known-good run, including the relevant commit, pipeline configuration, environment, and failure stage. Explain how you would determine whether the change exposed a code defect or a pipeline or environment issue.

How would you isolate an API failure in a cloud deployment?

What to coverSuggested preparation: Work through the request path and distinguish application errors from authentication, networking, configuration, or third-party API failures. Describe what evidence you would collect before recommending a workaround or escalating.

How would you investigate a customer issue when the relevant code spans unfamiliar languages?

What to coverSuggested preparation: Explain how you would follow inputs, outputs, error handling, and calls across code paths without assuming fluency in every language. Identify what you would verify with a small reproduction or targeted test before stating a cause.

What would make a technical escalation to engineering actionable?

What to coverSuggested preparation: Include concise reproduction steps, expected and actual behavior, relevant logs, environment details, and a hypothesis with supporting evidence. Separate confirmed findings from open questions so engineering can continue the investigation.

How would you turn a recurring support failure into a useful playbook or automation?

What to coverSuggested preparation: Explain how you would identify the repeated pattern and decide whether documentation, a diagnostic script, or another tool would reduce investigation time. Address how you would validate that it helps without obscuring unusual cases.

How would you plan release testing across different languages, IDEs, and development workflows?

What to coverSuggested preparation: Describe a risk-based set of representative environments and workflows, including checks for regressions and a way to record reproducible failures. Explain the trade-off between broad coverage and deeper testing of commonly affected paths.

How would you investigate an issue in an AI-assisted development workflow involving Devin or Windsurf?

What to coverSuggested preparation: Use the posting’s description of Devin as an AI software engineer and Windsurf as an AI-native IDE, without assuming undocumented product internals. Explain how you would separate an IDE or workflow configuration problem from a reproducible product behavior and communicate what remains uncertain.

Behavioural practice

Tell me about a time you resolved a technical issue despite an ambiguous initial report.

What to coverSuggested preparation: Use a real situation/action/result example showing how you gathered missing context, tested hypotheses, and reached or narrowed a resolution. Be clear about what you knew at each stage and what the outcome was.

Describe a time you managed several urgent technical issues at once.

What to coverSuggested preparation: Use a real situation/action/result example that explains how you prioritized, context-switched, and maintained investigation quality. Include how you kept affected people informed and what resulted.

Give an example of explaining a technical problem or workaround to a customer.

What to coverSuggested preparation: Set out the situation, the action you took to make the explanation accurate and usable, and the result. Distinguish a verified workaround from an unconfirmed root cause.

Tell me about a technical issue you escalated and worked through with engineering.

What to coverSuggested preparation: Use a real situation/action/result example covering the evidence you supplied, how you collaborated during the investigation, and the outcome. Clarify your own contribution without claiming certainty you did not have.

Describe a time you spotted a recurring failure and helped prevent repeat issues.

What to coverSuggested preparation: Explain the situation, how you identified the pattern, and what action you took through documentation, tooling, or feedback to another team. State the observed result rather than assuming the change eliminated every recurrence.

Sharpen your knowledge

Topics to prepare

Hypothesis-driven root-cause analysis
The posting requires debugging incomplete reports and isolating failures. Exercise: take an ambiguous issue description, list competing causes, then specify one observation that would confirm or rule out each.
Multi-component failure paths
The role calls for correlating logs, traces, errors, and code paths across systems. Exercise: sketch a request moving through an IDE, API, and service; mark where evidence could distinguish a client, network, or service failure.
Development-environment fundamentals
Working knowledge of Linux, Docker, Git, CI/CD, cloud platforms, and networking is required. Exercise: for a build that works locally but fails in CI, map how each layer could account for the difference and what evidence to seek.
Customer and engineering communication
The posting emphasizes clear customer updates and escalations with complete technical context. Exercise: outline what belongs in a customer-facing progress update versus an engineering handoff for the same unresolved issue.
AI-assisted development workflows
Cognition supports customers using Devin and Windsurf; familiarity with LLMs, AI-assisted tools, and IDEs is listed as nice to have. As a preparation suggestion, trace a typical code-editing workflow and identify where an environment-specific issue might surface; do not assume particular product features.

Skills to revise

Reproduce an ambiguous report
The role explicitly requires reproducing customer issues. Exercise: turn a vague 'fails in one environment' report into clarifying questions, recorded environment details, minimal reproduction steps, and an expected-versus-actual result.
Trace evidence to a failure boundary
The posting calls for log and trace analysis and reading code in languages such as Python, TypeScript, Java, or Go. Exercise: use a small application with an API error to follow the relevant code path, align log events, and state the narrowest supported failure boundary.
Inspect a failing development workflow
Linux, Docker, Git, CI/CD, networking, and cloud knowledge are explicit qualifications. Exercise: recreate a local-versus-CI failure in a small containerized repository; compare configuration, revision, connectivity, and error output before proposing a fix.
Write an actionable escalation
Engineering handoffs must include reproduction steps, logs, environment details, and root-cause hypotheses. Exercise: produce a concise escalation that separates observed facts from hypotheses and notes what remains unverified.
Turn a resolution into reusable guidance
The responsibilities include customer workarounds, internal playbooks, recurring-failure feedback, and release testing. Exercise: from a resolved issue, write a customer-safe workaround, a repeatable investigation checklist, and one regression test scenario.

Your preparation checklist

Map evidence to the stated qualifications
Prepare a brief inventory of relevant technical roles, education or equivalent practical experience, and examples involving Linux, Docker, Git, CI/CD, cloud platforms, networking, and reading code. The posting specifies 4+ years in a technical role and these areas of working knowledge.
Build a complex-issue walkthrough
Choose a real, anonymized example of an ambiguous development-environment issue. Show the initial report, reproduction, logs or traces examined, code paths followed, competing hypotheses, isolated cause, and resolution.
Practice a cross-stack debugging exercise
As preparation for the posting’s multi-component investigations, work through a hypothetical failure spanning an API, container, CI pipeline, and cloud environment. Produce a short diagnostic plan stating what evidence would distinguish application, configuration, and networking causes.
Draft an engineering escalation and customer update
From the same example, write two concise artifacts: an escalation with reproduction steps, environment details, relevant logs, findings, and open hypotheses; and a customer update that distinguishes confirmed facts from a possible workaround.
Show how you manage incoming issues
Prepare an example of prioritizing and context-switching across multiple technical issues while preserving investigation quality and clear communication. The posting emphasizes urgency and a high volume of incoming issues.
Prepare a reusable quality improvement
If you have a relevant example, bring a playbook, script, or documentation sample that made debugging or QA more systematic. Alternatively, sketch a proposed checklist for testing a release across languages, IDEs, and development workflows; present it as a preparation idea, not a Cognition process.
Write focused questions for Cognition
Ask how support engineers investigate Devin versus Windsurf issues, what context makes an engineering escalation useful, and how recurring failures inform product feedback and release testing. These questions follow responsibilities stated in the posting.

Availability last checked: 4 Nov 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in Singapore. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…