New opportunity

Advanced AI Full Stack Engineer

Accenture · Singapore

About this role

Accenture seeks an Advanced AI Full Stack Engineer in Singapore to design, build, deploy and maintain AI-powered applications across user interfaces, backend services and machine learning models. The role focuses on production systems powered by large language models (LLMs) and foundation models.

Responsibilities

Develop and optimize prompts and AI applications; architect generative agent systems and coordinate models to address complex tasks.
Define and evaluate AI architectures, improve model performance, accuracy and alignment, and integrate AI systems with products, APIs and tools.
Develop application and component strategies spanning user experience and backend systems. Build scalable ML infrastructure, including distributed training pipelines.
Evaluate robustness, safety and fairness, including hallucination risks. Review code, guide engineering improvements, troubleshoot testing and production issues, collaborate with business and technical teams, and document architectures and methods.

Requirements

The posting’s structured experience field specifies 120–144 months. The qualifications separately require at least 4 years designing and building resilient production AI systems, including controls, guardrails, context engineering, evaluations, CI/CD, tracing, logging and unit testing; at least 4 years creating AI solutions using patterns such as RAG and handling data pipelines; and at least 4 years in engineering teams using programming languages and frameworks such as Python, JavaScript, Java, Spring or Go, with mastery of an AI framework such as AutoGen, LangGraph or Semantic Kernel.
At least 4 years using application services on a public cloud such as AWS, GCP or Azure, including AI and generative AI capabilities; at least 3 years building AI applications or products, with knowledge of fine-tuning, prompt engineering, token utilization and latency optimization; and at least 2 years leading or helping manage a team.
Bachelor’s degree or equivalent minimum 12 years’ work experience; with an associate’s degree, a minimum of 6 years’ work experience.

Preferred

Experience training and evaluating LLMs and multimodal models; AI cost, ROI and tradeoff analysis; Model Context Protocol and agent-to-agent protocols; and mobile and web application development. Travel may range from 0% to 100%, depending on business and client needs.

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

Skills for this role

Large Language ModelsGenerative AIAI agentsRAGPrompt engineeringContext engineeringModel evaluationFine-tuningPythonJavaScriptJavaSpringGoAutoGenLangGraphSemantic KernelLangChainOpenAIAWSGoogle Cloud PlatformAzureCI/CDDistributed trainingAPI integrationTracingLoggingUnit testingCode reviewTeam leadershipModel Context Protocol

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

Deliver production AI applications end to end
Posting: The role spans user interfaces, backend services, models, deployment and maintenance; qualifications emphasize resilient systems, CI/CD, tracing, logging and unit tests. Preparation: Be ready to explain architectural decisions and how an AI application would be released, observed and maintained.
Design and optimize LLM and agent systems
Posting: Responsibilities include prompt optimization, agent systems with multi-model coordination, and evaluating AI architectures; qualifications cite RAG, context engineering, data pipelines, token use and latency. Preparation: Explain when you would use retrieval, agents or a simpler approach, and how you would measure the result.
Evaluate responsible AI behavior
Posting: The engineer must evaluate robustness, safety and fairness, minimize hallucination risks, and improve accuracy and alignment. Preparation: Outline evaluation criteria, guardrails and how findings would drive changes before and after release.
Build scalable integrations and ML infrastructure
Posting: Duties include distributed training pipelines and integration with APIs and tools; qualifications call for public-cloud application services. Preparation: Discuss scaling and integration choices, including where operational complexity is justified.
Lead engineering quality across teams
Posting: The role calls for code reviews, technical guidance, troubleshooting, documentation and collaboration with business leaders, engineers, architects and designers; qualifications include team leadership or management experience. Preparation: Show how you would communicate technical trade-offs and coordinate issue resolution across functions.

Likely challenges to think through

Balancing AI quality with responsiveness and cost
Likely challenge (inference): More retrieval, model calls or agent steps may improve task performance while increasing latency, token use and expense. Prepare to describe a measurement-led way to compare architectures against quality and performance targets.
Keeping agent behavior reliable in production
Likely challenge (inference): Multi-model coordination and tool integrations can make failures harder to isolate, while the posting stresses guardrails, evaluations and production troubleshooting. Prepare a failure-analysis approach using traces, logs, tests and targeted evaluations.
Reconciling model work with full-stack delivery
Likely challenge (inference): Work across model behavior, pipelines, APIs, backend systems and user experience may create competing priorities. Prepare to explain how you would define component boundaries and sequence work with business and technical stakeholders.

Practice your answers

Technical practice

How would you architect an LLM application from user interface to model and backend services?

What to coverThe role spans frontend experience, backend logic, models and production deployment. Suggested preparation: explain component boundaries, data flow, failure handling and how you would evaluate the architecture against user and business needs.

How would you design and evaluate a RAG system and its data pipeline?

What to coverThe qualifications call for RAG design patterns and data-pipeline experience. Suggested preparation: cover ingestion, retrieval, context selection and evidence-grounded evaluation, including what you would change when answers are inaccurate.

How would you improve prompts and context for a production AI application?

What to coverThe posting emphasizes prompt optimization, context engineering and resilient production systems. Suggested preparation: describe a repeatable evaluation approach, guardrails and trade-offs involving context size, token use and answer quality.

When would you use a generative agent system, and how would you coordinate its models and tools?

What to coverThe responsibilities include agent systems for complex tasks and multi-model coordination. Suggested preparation: explain orchestration, tool boundaries, failure recovery and how you would test whether agents improve on a simpler design.

How would you evaluate an LLM application for robustness, safety, fairness and hallucinations?

What to coverThe role explicitly calls for advanced evaluation across these risks. Suggested preparation: discuss representative test cases, measurable outcomes, review of failures and how findings would inform controls or model changes.

How would you decide between prompt changes and fine-tuning, and support training at scale?

What to coverThe posting covers fine-tuning foundation models and maintaining scalable infrastructure, including distributed training pipelines. Suggested preparation: compare expected gains, data needs and operational cost, then explain how you would validate and maintain the resulting pipeline.

How would you reduce token use and latency without sacrificing answer quality?

What to coverThe qualifications explicitly mention token utilization and latency optimization alongside model performance and accuracy. Suggested preparation: identify likely bottlenecks, compare optimization options and describe measurements that would reveal quality regressions.

How would you deploy and update an AI application using public-cloud services?

What to coverThe posting calls for public-cloud application services, AI capabilities and CI/CD practices. Suggested preparation: outline deployment boundaries, automated checks, release controls and how you would assess reliability after a change.

How would you diagnose a failure that appears only in production?

What to coverThe responsibilities include testing and production troubleshooting, while the qualifications mention tracing, logging and unit testing. Suggested preparation: explain how you would narrow the failure across application, retrieval, model and external integrations, then verify a fix.

What would you look for when reviewing code for an AI application?

What to coverThe role includes code reviews, engineering guidance and adherence to standards. Suggested preparation: cover testability, error handling, API integration, observability and whether changes preserve documented architecture and evaluation results.

Behavioural practice

Tell me about a time you guided an engineering team through a difficult technical decision.

What to coverThe posting includes team leadership, code reviews and guidance on improvements. Suggested preparation: use a real situation, your actions and the result, showing how you weighed alternatives and handled disagreement.

Describe a time you aligned an AI solution with different business and technical priorities.

What to coverThe role calls for collaboration with business leaders, engineers, architects and designers. Suggested preparation: give a real situation, explain how you translated competing objectives into a decision, and state the result.

Tell me about a production issue you helped resolve.

What to coverProduction troubleshooting is an explicit responsibility. Suggested preparation: describe the situation, your diagnostic and coordination actions, the result, and what changed to reduce recurrence.

Describe a time you raised a responsible-AI concern that affected a project decision.

What to coverThe posting emphasizes safety, fairness, robustness and hallucination risk. Suggested preparation: use a real example to explain the risk you identified, the evidence and actions you brought forward, and the outcome.

Tell me about a time you documented a complex system for others to maintain or extend.

What to coverThe role includes documenting architecture and methods for knowledge transfer. Suggested preparation: describe the audience and situation, what you documented, and the resulting effect on others' ability to work with the system.

Sharpen your knowledge

Topics to prepare

Resilient production AI architecture
Sketch an LLM application's request path and mark where context engineering, controls, guardrails, CI/CD, tracing, logging and unit tests fit. These are explicit production-system qualifications in the posting.
RAG and data-pipeline design
Diagram how source data becomes retrievable context, then identify failure points that could produce incomplete or unsupported answers. The posting explicitly requires experience with RAG patterns and data pipelines.
Agent and model coordination
For a complex task, compare a single-model workflow with an agent workflow that coordinates models; explain how you would limit and evaluate each step. Architecting generative agent systems is a stated responsibility.
Evaluation and responsible AI
Draft an evaluation plan covering accuracy, robustness, safety, fairness and hallucination risk, including what evidence would trigger a change. The role explicitly calls for these evaluations.
LLM performance tradeoffs
Work through a prompt-design decision involving token use, latency, accuracy and possible fine-tuning; explain how you would measure the tradeoff. Prompt engineering, fine-tuning knowledge and performance optimization appear in the posting.
Full-stack and cloud architecture
Draw the boundaries between a user interface, backend service, model, public-cloud application service and API integration; note where a distributed training pipeline would fit if needed. This reflects the advertised end-to-end scope and infrastructure responsibility.

Skills to revise

Implement and test a RAG flow
As a preparation exercise, build a small retrieval-to-answer flow with a data-ingestion step and tests for missing or irrelevant context. This practices the posting's RAG, data-pipeline and unit-testing requirements; the exercise is not an employer-prescribed project.
Prototype an agent workflow
Using one framework named in the posting, such as LangGraph, AutoGen or Semantic Kernel, prototype a two-step agent workflow and inspect its intermediate outputs. Framework mastery and agent-system implementation are explicitly relevant; no particular framework is mandated.
Integrate an AI service end to end
As a suggested exercise, connect a simple user interface to backend logic and an AI service on one public cloud named in the posting; document API boundaries and failure handling. The role spans frontend, backend, models and cloud services.
Instrument and evaluate a deployment
Add tracing, logging, unit tests and a small evaluation set to an AI application; rehearse diagnosing one poor answer and one latency issue. These practices match the stated production controls, evaluations and troubleshooting duties.
Review and explain engineering decisions
Practice reviewing a short AI-application change for correctness, guardrails and maintainability, then write a brief architecture decision and explain it to technical and business audiences. Code review, documentation, cross-functional collaboration and team leadership are explicit parts of the role.

Your preparation checklist

Map your evidence to the qualifications
Prepare a one-page matrix of relevant projects, dates and responsibilities against the posting’s AI engineering, RAG and data pipelines, programming and AI frameworks, public cloud, product development, and team leadership requirements. Note that the structured experience field says 120–144 months, while individual qualifications specify different minimums.
Build an end-to-end project walkthrough
Prepare a diagram and brief narrative for a production AI application, if you have one, tracing the user interface, backend, APIs, data pipeline, model integration and deployment. Explain your decisions and measured outcomes without implying experience you cannot substantiate.
Practice an LLM architecture exercise
Sketch a proposed RAG or agent-based system using a relevant framework such as LangGraph, AutoGen or Semantic Kernel. Be ready to explain context engineering, prompt design, tool or model coordination, and tradeoffs in token use and latency; this is preparation, not a stated interview exercise.
Prepare an evaluation and safety plan
Draft sample evaluation cases and measures for accuracy, robustness, hallucinations, safety and fairness. Include how guardrails, tracing, logging and unit tests would help detect failures and guide improvements.
Explain production and infrastructure choices
Prepare a concise account of cloud services, CI/CD, incident troubleshooting and API integration from a relevant project. Review how you would approach scalable ML infrastructure or distributed training pipelines, which the posting names as responsibilities.
Select a leadership and collaboration example
Prepare one specific example of code review, technical guidance or issue resolution, including how you worked with engineering and business stakeholders and documented decisions. The posting also specifies team leadership or management experience.
Write questions for the employer
Ask which systems this role would own, how the team evaluates production LLM quality, how frontend and backend work is divided, and whether distributed training is in scope. Clarify expected travel, which the posting says can vary with client needs.

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…