New opportunity

AI / ML Engineer

Accenture · Bengaluru, India

About this role

Accenture seeks an AI / ML Engineer in Bengaluru to develop and deploy production-quality AI applications, services and agentic workflows for enterprise and client environments. The role uses AI tools and cloud AI services within cloud or on-premises application pipelines, including generative AI models.

Responsibilities

Build AI services, agents, retrieval pipelines, tools and application integrations; develop Python services and APIs and connect them to enterprise data sources and business applications.
Implement automated tests, evaluations, logging, tracing and monitoring. Troubleshoot model outputs, data quality, integration failures and performance issues.
Support deployment, user testing and production stabilization, and document reusable components and implementation patterns.

Requirements

Large Language Models (LLMs) are a must-have skill. Apply Python, APIs, data structures, software engineering and testing, alongside RAG, embeddings, vector databases, tool calling and prompt design.
Work with cloud services, Docker, Git, CI/CD, SQL and backend frameworks, with attention to secure coding, observability, latency and cost.
Have contributed meaningful code to production software, AI, ML or data systems; coursework, certifications and prototype-only experience are insufficient. Have practical experience using Cursor, GitHub Copilot, Claude Code, Codex or Replit Agent to build and deploy AI services, including specifying tasks, validating generated code, creating tests, troubleshooting failures and working safely with Git and CI/CD.
The posting states a minimum of 3 years of experience and 15 years of full-time education; its job header lists an experience range of 2–5 years. The position is full-time.

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

Skills for this role

Large Language Models (LLMs)PythonAPIsRetrieval-Augmented Generation (RAG)EmbeddingsVector databasesTool callingPrompt designCloud servicesDockerGitCI/CDSQLBackend frameworksAutomated testingAI evaluationLoggingTracingMonitoringSecure codingCursorGitHub CopilotClaude CodeCodexReplit Agent

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-ready AI services
The role calls for Python services and APIs that deploy AI applications and agentic workflows through cloud or on-premises pipelines, not prototype-only work.
Build LLM and retrieval integrations
LLMs are the must-have skill. The work includes agents, retrieval pipelines, tools, and connections to enterprise data sources and business applications; RAG, embeddings, vector databases, and tool calling are named technical skills.
Make systems testable and observable
The posting assigns automated tests, evaluations, logging, tracing, and monitoring, alongside troubleshooting model outputs, data quality, integration failures, and performance.
Support release and reuse
Responsibilities extend to deployment, user testing, production stabilization, and documenting reusable components. The posting also emphasizes validating AI-generated code, creating tests, and working safely with Git and CI/CD.

Likely challenges to think through

Balancing answer quality with operating constraints
Preparation hypothesis: An LLM or retrieval pipeline may need better outputs without unacceptable latency or cost. Prepare to explain how you would evaluate output quality, trace failures, and measure performance before changing retrieval, prompts, or model usage.
Integrating enterprise data safely
Preparation hypothesis: Connecting AI services to data sources and business applications can put delivery speed in tension with secure coding and data quality. Prepare an example of how you would validate inputs, handle integration failures, and test API behavior.
Keeping AI-assisted development production-safe
Preparation hypothesis: Coding agents may accelerate implementation while introducing defects that surface during deployment or user testing. Prepare to describe task specifications, review and tests for generated code, and safe Git and CI/CD practices during stabilization.

Practice your answers

Technical practice

How would you design a production-quality Python API for an LLM-backed service?

What to coverThe role calls for Python services, APIs and production-ready AI applications. Suggested preparation: explain request validation, error handling, testing and secure integration with a business application.

How would you build a retrieval pipeline over enterprise data?

What to coverThe posting names RAG, embeddings, vector databases and enterprise data sources. Suggested preparation: discuss preparing data, retrieving relevant material and checking whether retrieved content supports the answer.

When would an agent use a tool rather than generate an answer directly?

What to coverThe role includes agents, tool calling and application integrations. Suggested preparation: explain tool selection, input and output validation, and how to handle a failed or inappropriate tool call.

How would you diagnose an inaccurate LLM response?

What to coverThe posting calls for prompt design and troubleshooting model outputs. Suggested preparation: distinguish prompt, retrieval and model-output issues, then explain how you would test a proposed fix.

What automated tests would you add to an AI service?

What to coverAutomated testing is an explicit responsibility. Suggested preparation: distinguish deterministic tests for Python code and APIs from checks on variable model outputs, and explain what belongs in CI/CD.

How would you evaluate an AI application before and after deployment?

What to coverThe role explicitly includes evaluations, user testing and production stabilization. Suggested preparation: describe representative test cases, criteria tied to the application’s intended behavior, and how user feedback would inform changes.

How would you trace a failure across an AI service and its data integrations?

What to coverThe posting names logging, tracing, monitoring, data-quality problems and integration failures. Suggested preparation: walk through isolating the failing stage while keeping sensitive data out of diagnostic records.

How would you deploy and maintain an AI service in a cloud or on-premises pipeline?

What to coverThe role mentions cloud or on-premises pipelines, Docker, Git and CI/CD. Suggested preparation: explain packaging, configuration, automated checks, deployment verification and a safe approach to reverting a faulty change.

How would you investigate high latency or cost in an LLM application?

What to coverLatency, cost and performance issues are stated considerations. Suggested preparation: describe what you would measure across retrieval, model calls and integrations, then discuss trade-offs before changing the design.

How would you use an AI coding assistant safely to deliver an AI service?

What to coverThe posting requests practical use of tools such as Cursor, GitHub Copilot, Claude Code, Codex or Replit Agent. Suggested preparation: explain task specifications, review of generated code, tests, failure diagnosis and safe Git and CI/CD practices.

Behavioural practice

Tell me about a meaningful code contribution to a production system.

What to coverThe posting distinguishes production contributions from coursework and prototypes. Suggested preparation: use a real situation, your specific coding actions and the resulting production outcome; do not substitute a hypothetical project.

Describe a time you investigated a difficult production failure.

What to coverTroubleshooting model outputs, data quality, integrations and performance is part of the role. Suggested preparation: give the situation, how you narrowed down the cause, what you changed and the observed result.

Tell me about a time user testing changed an implementation.

What to coverThe role includes user testing and production stabilization. Suggested preparation: describe the feedback situation, your response and the result, distinguishing observed behavior from assumptions.

Describe how you handled a questionable change proposed by an AI coding assistant.

What to coverThe posting emphasizes validating generated code, troubleshooting failures and working safely with Git and CI/CD. Suggested preparation: if you have a real example, explain the situation, your verification and corrective actions, and the outcome.

Tell me about a reusable component or implementation pattern you documented.

What to coverDocumenting reusable components and patterns is a stated responsibility. Suggested preparation: describe the need, what you built or documented, and the result for people using or maintaining it.

Sharpen your knowledge

Topics to prepare

LLM behavior and agentic workflows
The posting names LLMs as the must-have skill and calls for agents, tool calling and prompt design. Suggested revision exercise: trace an agent request from prompt to tool call to final answer, identifying where an incorrect output or unsafe tool action could arise.
Retrieval-augmented generation
The posting explicitly names RAG, embeddings, vector databases and retrieval pipelines. Suggested revision exercise: diagram how source data becomes indexed content and retrieved context, then explain how you would investigate an irrelevant retrieval result.
Production AI architecture
The role develops production-quality AI applications using cloud AI services in cloud or on-premises pipelines. Suggested revision exercise: sketch a service-to-model-to-data-source pipeline and explain its integration, latency, cost and security trade-offs.
Evaluation and observability
The responsibilities include evaluations, logging, tracing and monitoring, plus troubleshooting model outputs and performance. Suggested revision exercise: define a small set of test inputs, expected behaviors and diagnostic signals for an AI service failure.
Backend and data fundamentals
The posting lists APIs, data structures, SQL and backend frameworks and calls for enterprise-data integration. Suggested revision exercise: explain an API request’s data flow through validation, a SQL query and a response, including likely failure points.

Skills to revise

Build a Python AI-service endpoint
The posting calls for production-quality Python services and APIs. Suggested hands-on exercise: implement a small endpoint with input validation, error handling and a replaceable model-call interface; explain the choices that make it maintainable.
Implement and inspect a retrieval pipeline
Building retrieval pipelines is an explicit responsibility. Suggested hands-on exercise: create a small RAG flow with embeddings and a vector database, then inspect retrieved results for an unanswered or poorly answered query.
Test an agent’s tool calls
The role builds agents and tools and requires automated tests. Suggested hands-on exercise: implement one constrained tool call and tests for valid input, rejected input and tool failure; check that the resulting response reflects the outcome.
Diagnose an integration failure
The posting calls for troubleshooting data quality, integration failures and model outputs. Suggested hands-on exercise: introduce a malformed data response, use logs and a trace to locate the failure, then add a regression test.
Use an AI coding assistant with review controls
The posting requires practical use of Cursor, GitHub Copilot, Claude Code, Codex or Replit Agent, including task specifications and generated-code validation. Suggested hands-on exercise: use one listed assistant to propose a small change, review its diff, run tests and commit it safely with Git and CI/CD checks.
Package and stabilize a service
Docker, cloud services, deployment, user testing and production stabilization appear in the posting. Suggested hands-on exercise: containerize a small Python service, run a user-facing test, and document a startup or performance issue and its fix; no particular cloud provider is specified.

Your preparation checklist

Select a production-code walkthrough
Prepare one example of meaningful code contributed to a production software, AI, ML or data system. Show the problem, your code changes, tests, deployment path and outcome; the posting says prototypes alone are insufficient.
Practice an LLM system design
Sketch a hypothetical enterprise AI service using retrieval, embeddings, a vector database and tool calling. Explain data-source integration, prompt design, security, latency and cost trade-offs as design choices, not as a claimed Accenture architecture.
Prepare a Python API walkthrough
Bring a concise diagram or code excerpt showing a Python service, its API contract and a connection to a data source or business application. Be ready to explain error handling and automated tests.
Demonstrate safe use of an AI coding tool
Prepare an example using a tool named in the posting, if you have one: show the task specification, how you validated generated code, tests you added, a failure you corrected and how you used Git and CI/CD safely.
Build an evaluation and troubleshooting plan
Draft a small test and evaluation matrix for model outputs, retrieval quality, integration failures and performance. Specify what logging, tracing and monitoring would help diagnose each issue.
Rehearse deployment and stabilization decisions
Outline how you would package and release an AI service, support user testing, handle a production issue and document a reusable component. Relate the plan to the posting’s cloud or on-premises pipelines, Docker and CI/CD.
Write targeted questions for the employer
Ask which AI services and integrations this role would own, how production evaluations and monitoring are handled, and how user testing feeds into stabilization. You can also clarify the page’s AI / ML Engineer project role versus its AI Infrastructure Architecture Senior Analyst header.

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 India. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…