New opportunity

Large Language Model Architect

Accenture · Pune, India

About this role

Accenture is seeking a Large Language Model Architect for a full-time position based at its Pune office. The role focuses on designing large language models that process and generate natural language, including defining neural network parameters and working with models trained on large quantities of unlabeled text.

Responsibilities

Collaborate with cross-functional teams to define model parameters, experiment with neural network architectures, analyze model performance, and troubleshoot and optimize models for application and user needs.
Serve as a subject-matter expert, manage and mentor team members, make team decisions, contribute to decisions across multiple teams, and solve problems for the immediate team and other teams.
Evaluate and implement best practices in model architecture and deployment.

Requirements

Proficiency in AI Agents & Workflow Integration is required. The posting lists 5–10 years of experience for the position and separately specifies a minimum of 7.5 years in AI Agents & Workflow Integration.
Strong understanding of neural network design and optimization; experience preprocessing data and handling large datasets for model training; and familiarity with cloud-based AI model and service deployment.
Fifteen years of full-time education is required.

Preferred skills include Python, API Management, API security, generative AI, large language models, retrieval-augmented generation, AWS AI Services, Microsoft Azure AI Foundry, Microsoft Azure Machine Learning, and Google Cloud Vertex AI services.

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

Skills for this role

AI Agents & Workflow IntegrationLarge Language Models (LLMs)Neural network designModel optimizationNatural language processingData preprocessingPythonAPI ManagementAPI securityGenerative AIRetrieval-Augmented GenerationAWS AI ServicesMicrosoft Azure AI FoundryMicrosoft Azure Machine LearningGoogle Cloud Vertex AIVertex AI VisionCloud-based AI deploymentTeam leadershipMentoring

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

AI agents and workflow integration
The posting names this as the must-have skill and separately specifies at least 7.5 years of experience in it. Prepare to explain how agent workflows connect model capabilities to application needs.
LLM architecture and performance
The role involves defining neural-network parameters, experimenting with architectures, analyzing results, and troubleshooting and optimizing models that process and generate language.
Technical leadership across teams
The architect is expected to act as a subject-matter expert, manage and mentor team members, make team decisions, and contribute to decisions across multiple teams.
Data and deployment practices
The posting calls for experience preprocessing large training datasets, familiarity with cloud-based AI deployment, and ongoing evaluation of architecture and deployment best practices.

Likely challenges to think through

Balancing model quality with application needs
A likely challenge is deciding which architecture or optimization changes improve results without losing sight of varied user requirements. Prepare an example of how you measured performance, investigated a shortcoming, and chose a trade-off.
Connecting agent workflows to deployable systems
Because agent integration is required and cloud deployment and API-related skills are mentioned, integration may involve reliability and security trade-offs. Prepare to describe design choices and safeguards without assuming a particular platform or architecture is used.
Aligning decisions across teams
Cross-team decisions may require reconciling competing technical priorities while mentoring the immediate team. Prepare an example showing how you made a decision, communicated its rationale, and helped others carry it forward.

Practice your answers

Technical practice

How would you design an AI agent workflow that integrates an LLM with application services?

What to coverAI Agents & Workflow Integration is the posting's required skill. Suggested preparation: Explain how you would define the agent's steps, connect services, handle failures, and evaluate whether the workflow meets user needs.

How would you choose neural network architecture and model parameters for a natural-language task?

What to coverThe role involves defining parameters and experimenting with architectures for models that process and generate language. Suggested preparation: Discuss task requirements, model capacity, compute constraints, and how you would compare alternatives.

How would you prepare a large collection of unlabeled text for model training?

What to coverThe posting calls for data preprocessing and handling large training datasets. Suggested preparation: Cover data quality, deduplication, representation, and checks that prevent unsuitable data from undermining training.

How would you evaluate whether an LLM experiment improves performance?

What to coverAnalyzing model performance is part of the advertised work. Suggested preparation: Describe task-relevant evaluation, consistent comparisons, error analysis, and the trade-off between quality and resource use.

How would you diagnose and optimize an LLM that performs poorly for an application?

What to coverThe role includes troubleshooting and optimizing models for application and user needs. Suggested preparation: Explain how you would isolate data, model, and integration issues before choosing and validating a change.

When would you use retrieval-augmented generation rather than relying on a model alone?

What to coverRetrieval-augmented generation is listed as a good-to-have skill. Suggested preparation: Compare access to relevant information with added retrieval complexity, and explain how you would assess the combined system's answers.

How would you manage and secure an API that exposes an LLM-backed capability?

What to coverAPI Management and API security appear among the good-to-have skills. Suggested preparation: Discuss access control, protection of sensitive data, operational limits, and how these choices affect integration.

What would you consider when deploying an AI model as a cloud-based service?

What to coverThe posting asks for familiarity with cloud-based AI deployment and lists several cloud AI services as good-to-have skills. Suggested preparation: Compare deployment options using performance, scalability, cost, monitoring, and maintainability without assuming a mandated platform.

How would you decide whether a proposed model architecture is suitable for different user needs?

What to coverThe role calls for solutions that meet diverse application and user requirements. Suggested preparation: Describe how you would translate requirements into testable criteria and weigh quality against practical constraints.

Behavioural practice

Tell me about a time you aligned cross-functional contributors on a technical approach.

What to coverThe posting emphasizes collaboration across functions to define model parameters and analyze results. Suggested preparation: Give a real situation, your actions to resolve differing priorities, and the resulting decision or outcome.

Describe a consequential technical decision you made while managing a team.

What to coverThe role includes managing team performance and responsibility for team decisions. Suggested preparation: Explain the situation, how you weighed evidence and involved the team, and what resulted.

Tell me about a problem you helped solve across multiple teams.

What to coverThe posting expects contributions to key decisions and solutions beyond the immediate team. Suggested preparation: Identify the situation, your specific actions across team boundaries, and the result.

Describe a time you mentored a junior colleague on an AI-related topic.

What to coverMentoring junior team members in AI technologies is an advertised responsibility. Suggested preparation: Give a real situation, how you adapted your guidance, and the observable result without overstating your contribution.

Tell me about a time you improved a model architecture or deployment practice.

What to coverThe role calls for continually evaluating and implementing best practices in architecture and deployment. Suggested preparation: Describe the situation, the evidence behind your proposed change, your actions, and the result.

Sharpen your knowledge

Topics to prepare

AI agent and workflow architecture
Posting: AI Agents & Workflow Integration is the must-have skill. Exercise: sketch how an agent passes a user request through a workflow, identifying decision points, handoffs, and failure paths; explain the design choices.
LLM and neural network design
Posting: the role designs models that process and generate natural language and defines neural network parameters. Exercise: compare two architecture or parameter choices and explain their expected effect on model behavior.
Model evaluation and optimization
Posting: the architect analyzes, troubleshoots, and optimizes model performance for application and user needs. Exercise: take a hypothetical quality-versus-latency trade-off and explain what you would measure before changing the model.
Large-scale training data
Posting: experience preprocessing data and handling large datasets for training is specified; the role description mentions large quantities of unlabeled text. Exercise: outline how you would check data quality and prepare text before training.
Deployment and API considerations
Posting: cloud-based AI deployment familiarity is specified; API Management and API security are good-to-have skills. Exercise: explain the design considerations for exposing a deployed model through an API, including access control and operational trade-offs.

Skills to revise

Map an agent workflow
Posting: agent and workflow integration is required. Exercise: draw a request-to-response workflow with inputs, outputs, handoffs, and recovery steps, then practice explaining how you would integrate it with an application.
Preprocess a text dataset
Posting: data preprocessing and large-dataset handling are specified. Exercise: use a small sample to identify duplicates, malformed records, and unsuitable text; document how the same checks could scale.
Diagnose model results
Posting: performance analysis, troubleshooting, and optimization are responsibilities. Exercise: review a sample set of model responses, group the errors, and propose one measurable change to test first.
Prototype an API interaction in Python
Posting: Python and API Management are good-to-have skills. Exercise: as optional preparation, write a small Python request-and-response example that validates inputs and handles an API error; explain where access controls would apply.
Present an architecture decision
Posting: the architect makes team decisions, contributes across teams, and mentors junior members. Exercise: prepare a short decision brief comparing two model approaches, including evidence, trade-offs, and how you would explain the choice to another team.

Your preparation checklist

Map the stated requirements
The posting lists 5–10 years for the role, a minimum of 7.5 years in AI Agents & Workflow Integration, and 15 years of full-time education. Prepare a dated experience timeline and education record that address those distinct statements.
Build an agent-workflow walkthrough
AI Agents & Workflow Integration is the required skill. Prepare one project walkthrough showing the workflow design, integrations, decisions, failure handling, and measurable results; distinguish your own contributions from the team’s.
Practice an LLM architecture exercise
The role involves neural network parameters, architecture experiments, and natural-language processing and generation. Sketch a design for a stated use case, explaining parameter and architecture choices, trade-offs, and how you would test alternatives.
Document data and model-optimization decisions
The posting calls for preprocessing large training datasets and analyzing, troubleshooting, and optimizing model performance. Prepare an example with data checks, evaluation measures, a diagnosed issue, and the evidence behind an improvement.
Prepare a deployment design
Cloud-based AI deployment is listed, while API management, API security, RAG, and named cloud services are good-to-have skills. As a preparation exercise, diagram one deployment you know and be ready to explain its interfaces, security choices, and operational trade-offs without implying every listed service is required.
Show how you lead technical decisions
The role expects an SME who manages and mentors team members and contributes to decisions across teams. Prepare a specific example of a contested technical decision, how you aligned stakeholders, and what changed after mentoring or collaboration.
Write questions for the employer
Bring questions grounded in the posting: which agent workflows are in scope, how LLM performance is evaluated, what deployment environment the team uses, and how architectural decisions are shared across teams.

Availability last checked: 5 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…