Accenture seeks an AI / ML Engineer in Bengaluru to own NLP and generative AI solutions from design through production. The role focuses on scalable, cloud-native, LLM-powered applications on Azure, including chatbots, retrieval-augmented generation (RAG) systems and AI assistants.
Responsibilities
Design data ingestion, preprocessing and document-update pipelines; implement prompt engineering, embeddings, vector search and evaluation.
Build asynchronous REST APIs with FastAPI or Flask, and manage rate limits, retries, load balancing and fault tolerance for LLM APIs.
Deploy, monitor and scale applications on Azure; deliver proofs of concept and minimum viable products under tight timelines.
Mentor junior colleagues and support team delivery. The broader project role includes production-ready AI applications using cloud AI services or on-premises pipelines, potentially involving deep learning, neural networks, chatbots and image processing.
Requirements
Large language models are the required skill. The posting states a minimum of 7.5 years of experience and 15 years of full-time education. It also lists 5–10 years in the job header and 5+ years in NLP, AI or ML, along with proven production deployment of generative AI systems.
Advanced Python; NLP, generative AI, RAG and prompt engineering; Azure OpenAI or other LLM platforms; asynchronous FastAPI or Flask; and Azure services including App Service, Storage, AI Search and Cosmos DB.
CI/CD, logging and monitoring, including Jenkins and OpenTelemetry. Additional technologies listed include vector databases, agentic libraries such as LangChain, LangGraph and AutoGen, computer vision and document processing.
For full details about this role, click “Apply on company site” to view the employer’s original posting.
Skills for this role
PythonNatural Language ProcessingGenerative AILarge Language ModelsRetrieval-Augmented GenerationPrompt EngineeringEmbeddingsVector SearchVector DatabasesAzure OpenAIAzure CloudAzure App ServiceAzure StorageAzure AI SearchAzure Cosmos DBFastAPIFlaskREST APIsCI/CDJenkinsOpenTelemetryLoggingMonitoringLangChainLangGraphAutoGenComputer VisionDocument ProcessingData IngestionData Preprocessing
Your skill match
Checking your profile…
YOUR NEXT STEP
Get interview-ready for this role
A focused preparation guide, built around this job’s responsibilities and requirements.
✦ AI-generated guide
Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.
01
Understand the role
What matters most
End-to-End Generative AI Lifecycle Management
The role requires owning the full lifecycle of LLM-powered applications from initial design and data ingestion to production deployment on Azure. You must demonstrate proficiency in building robust RAG systems, managing document pipelines, and implementing advanced prompt engineering and evaluation frameworks.
Scalable Cloud-Native Architecture on Azure
You are expected to build and maintain production-ready, asynchronous REST APIs using FastAPI or Flask. This involves implementing critical infrastructure patterns such as load balancing, fault tolerance, rate limiting, and retry logic to ensure high availability when interacting with LLM APIs.
Technical Leadership and Delivery
Beyond individual development, the role demands delivering rapid PoCs and MVPs under tight deadlines. You will also be responsible for mentoring junior team members and ensuring the team meets delivery standards, requiring a balance between hands-on coding and technical oversight.
Likely challenges to think through
Balancing Rapid Prototyping with Production Stability
The requirement to deliver MVPs under tight timelines while maintaining production-ready quality creates a tension between speed and reliability. Prepare by discussing your strategies for building modular, scalable codebases that allow for quick iteration without sacrificing long-term maintainability or system robustness.
Managing LLM API Constraints and Reliability
Integrating LLMs into production environments introduces challenges related to latency, rate limits, and non-deterministic outputs. Prepare by detailing your experience with implementing asynchronous patterns, robust logging via OpenTelemetry, and defensive programming techniques to handle API failures and ensure consistent system performance.
03
Practice your answers
Technical practice
Designing RAG Pipelines for Document Updates
What to coverExplain how you would architect a data ingestion pipeline that handles frequent document updates while maintaining search accuracy. Discuss the trade-offs between full re-indexing versus incremental updates and how you manage vector database consistency.
Optimizing LLM API Resilience
What to coverDescribe your approach to implementing fault tolerance when consuming LLM APIs. Focus on strategies for handling rate limits, implementing exponential backoff, and using circuit breakers to maintain system stability during high-latency periods.
Evaluating Generative AI Output
What to coverWhat metrics and frameworks do you use to evaluate the quality of RAG-based systems? Discuss how you balance automated evaluation metrics with human-in-the-loop feedback to ensure production-ready performance.
Asynchronous API Development with FastAPI
What to coverWhy is asynchronous programming critical for LLM-powered REST APIs? Explain how you manage event loops and non-blocking I/O in FastAPI to handle concurrent requests effectively without exhausting system resources.
Agentic Workflows with LangGraph or AutoGen
What to coverCompare the use of LangChain versus LangGraph for building complex AI agents. Provide an example of how you would structure a multi-agent system to solve a specific business problem while managing state and tool execution.
Azure Cloud Infrastructure for AI
What to coverHow do you leverage Azure AI Search and Cosmos DB to support scalable vector search? Discuss the configuration considerations for indexing strategies and latency optimization in a production Azure environment.
Monitoring and Observability in GenAI
What to coverHow do you implement effective logging and monitoring for LLM applications using OpenTelemetry? Detail how you track token usage, latency, and prompt-response quality in a live production environment.
Prompt Engineering at Scale
What to coverDescribe your methodology for versioning and managing prompts in a production system. Suggestion: Discuss how you handle prompt drift and ensure consistency across different model versions or fine-tuned iterations.
Behavioural practice
Delivering Under Tight Timelines
What to coverDescribe a situation where you had to deliver a PoC or MVP under significant time pressure. Explain how you prioritized features, managed stakeholder expectations, and ensured the quality of the final deliverable.
Mentoring Junior Engineers
What to coverProvide an example of how you have supported the growth of junior team members. Focus on how you balanced your own delivery responsibilities with the need to provide technical guidance and code reviews.
Handling Production Failures
What to coverTell me about a time a production AI system failed or performed unexpectedly. Describe your troubleshooting process, how you identified the root cause, and the steps you took to prevent recurrence.
Navigating Technical Trade-offs
What to coverDescribe a scenario where you had to choose between two competing technical approaches for an AI solution. Explain how you evaluated the trade-offs, such as cost versus performance, and how you justified your decision to the team.
02
Sharpen your knowledge
Topics to prepare
Retrieval-Augmented Generation (RAG) Architecture
The role requires designing end-to-end RAG systems. Review the conceptual flow of data ingestion, document preprocessing, embedding generation, and vector search. Exercise: Sketch a system architecture diagram for a document-based Q&A bot, identifying where bottlenecks occur in retrieval and how to mitigate them using specific Azure AI Search configurations.
Production-Grade LLM API Management
The position emphasizes handling rate limits, retries, and fault tolerance for LLM-powered applications. Review strategies for implementing asynchronous patterns in production. Exercise: Design a robust error-handling strategy for an LLM API call that includes exponential backoff, circuit breaking, and fallback mechanisms to ensure system stability.
Agentic Workflow Design
The role involves using libraries like LangChain, LangGraph, and AutoGen to build AI assistants. Review the conceptual differences between simple chains and agentic workflows. Exercise: Compare the state management capabilities of LangGraph versus standard LangChain chains for a multi-step task requiring tool-use and memory.
Skills to revise
Asynchronous API Development with FastAPI
The role requires building asynchronous REST APIs. Exercise: Develop a FastAPI endpoint that performs an asynchronous call to an LLM service, incorporating proper logging and OpenTelemetry instrumentation to track request latency and success rates.
Azure Cloud AI Integration
The role requires deploying applications on Azure using App Service, Cosmos DB, and AI Search. Exercise: Create a configuration plan for deploying a containerized Python application to Azure App Service that connects to Azure AI Search for vector retrieval and Cosmos DB for session state management.
CI/CD and Monitoring Implementation
The role requires managing production deployments with Jenkins and OpenTelemetry. Exercise: Draft a Jenkins pipeline script that includes stages for automated testing, container image building, and deployment, and define a set of custom metrics to monitor via OpenTelemetry for an LLM-based application.
04
Your preparation checklist
Prepare a production-grade RAG architecture walkthrough
Develop a technical summary of a RAG system you have deployed, focusing on data ingestion pipelines, vector database selection, and handling document updates. Be prepared to explain how you managed retrieval accuracy and latency in a production environment.
Demonstrate asynchronous API design patterns
Create a code sample using FastAPI that demonstrates handling high-concurrency LLM requests. Include logic for implementing rate limiting, retries, and fault tolerance to ensure system stability when interacting with external LLM APIs.
Map Azure service expertise to LLM workflows
Prepare to discuss how you have integrated Azure AI Search, Cosmos DB, and App Service into an end-to-end AI solution. Specifically, explain your approach to monitoring these services using OpenTelemetry and logging for production observability.
Formulate strategies for agentic workflows
Review your experience with libraries like LangChain, LangGraph, or AutoGen. Prepare to explain how you have designed agentic systems to solve complex tasks and how you evaluate the performance of these agents in a real-world setting.
Draft technical questions for the interviewers
Prepare questions regarding the team's current CI/CD maturity for AI models, the specific challenges they face with LLM evaluation in production, and how the team balances rapid MVP delivery with long-term system scalability on Azure.
Availability last checked: 4 Nov 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy
Share this opportunityKnow someone who would be a great fit?