New opportunity

Python & GenAI Engineer

Capgemini · Bangalore, India

About this role

Capgemini seeks a Python & GenAI Engineer in Bangalore to develop and maintain generative AI applications and integrate AI capabilities into enterprise software. The posting describes the position as permanent, with support for hybrid working.

Responsibilities

Develop and enhance applications using large language models and AI frameworks. Build and optimize retrieval-augmented generation (RAG) pipelines, vector search and AI-powered workflows.
Design prompts and fine-tune models using techniques such as LoRA and QLoRA. Develop REST APIs to connect AI capabilities with enterprise applications.
Deploy, monitor and maintain AI solutions in cloud and containerized environments. Test, debug, document and review code, while following Responsible AI, security and data governance practices.

Profile:

Strong Python, object-oriented programming, API and debugging skills, with knowledge of software development best practices, data structures, algorithms, complexity analysis and problem-solving.
Hands-on experience with LLMs, prompt engineering, RAG, AI agents and embeddings. Experience with LangChain, LangGraph, LlamaIndex, Hugging Face or similar frameworks; knowledge of vector databases, semantic and hybrid search, and retrieval architectures.
Experience with PyTorch, TensorFlow or Keras; familiarity with Docker, Git, CI/CD and Azure, AWS or GCP. Understanding of AI governance, data privacy and Responsible AI principles.

Capgemini highlights career development and training and certification opportunities, including in Python and generative AI. No specific years of experience or salary are stated.

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

Skills for this role

PythonObject-Oriented ProgrammingREST APIsDebuggingData StructuresAlgorithmsComplexity AnalysisGenerative AILarge Language ModelsPrompt EngineeringRAGAI AgentsEmbeddingsLangChainLangGraphLlamaIndexHugging FaceVector DatabasesVector SearchSemantic SearchHybrid SearchLoRAQLoRAModel Fine-TuningPyTorchTensorFlowKerasDockerGitCI/CD

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

RAG and retrieval quality
The role builds and optimizes RAG pipelines, vector search, and AI-powered workflows. Prepare to explain how embeddings, vector databases, and semantic or hybrid search affect retrieval results.
LLM application design
The posting calls for LLM applications, effective prompts, and model fine-tuning with LoRA or QLoRA. Prepare to explain when you would adjust prompts, retrieval, or the model itself.
Enterprise integration
Developing REST APIs to connect AI capabilities with enterprise applications is an explicit duty. Prepare to discuss Python API design, error handling, debugging, and maintainable code.
Deployment and code quality
The role includes cloud and containerized deployment, monitoring, maintenance, testing, documentation, and code reviews. Prepare to walk through how an AI application moves from development into reliable operation.
Responsible AI and data protection
The posting explicitly requires Responsible AI, security, data governance, and an understanding of data privacy. Prepare to describe how these concerns shape an application’s design and review.

Likely challenges to think through

Retrieval quality versus response time
Preparation hypothesis: More extensive retrieval may improve answer grounding while increasing latency and complexity. Be ready to explain how you would evaluate retrieval results and tune a RAG pipeline for both relevance and responsiveness.
Choosing the right adaptation method
Preparation hypothesis: Prompt changes, retrieval improvements, and LoRA or QLoRA fine-tuning involve different effort and maintenance costs. Prepare a decision framework for diagnosing the problem before choosing an approach.
Useful answers within data boundaries
Preparation hypothesis: Enterprise AI integration may require balancing useful responses against privacy, security, and governance constraints. Prepare to explain where data controls, testing, and review would fit in the workflow.

Practice your answers

Technical practice

How would you structure a Python application that serves an LLM-powered workflow?

What to coverThe posting calls for strong Python, object-oriented programming, data structures, algorithms and complexity analysis. Suggested preparation: explain how you would separate components, choose data structures and assess performance as the workflow grows.

How would you build and evaluate a retrieval-augmented generation pipeline?

What to coverBuilding and optimizing RAG pipelines is a stated responsibility. Suggested preparation: walk through document preparation, embeddings, retrieval and answer generation, then explain how you would evaluate answer quality and identify whether a failure came from retrieval or generation.

When would you use semantic search, hybrid search or a different retrieval approach?

What to coverThe posting names vector databases, vector search, semantic and hybrid search, and retrieval architectures. Suggested preparation: compare their strengths for different query types and discuss relevance, latency and operational complexity when choosing an approach.

How would you improve a prompt that produces inconsistent answers?

What to coverEffective prompt design and prompt engineering are part of the role. Suggested preparation: describe controlled changes to instructions and context, how you would test them against representative inputs, and how you would check for regressions.

How would you decide whether to use prompting, LoRA or QLoRA for a model task?

What to coverThe role explicitly mentions prompt design and fine-tuning with LoRA and QLoRA, alongside experience with PyTorch, TensorFlow or Keras. Suggested preparation: compare the approaches in terms of task fit, data needs, compute constraints and evaluation before discussing implementation.

How would you design and troubleshoot an AI-agent workflow?

What to coverThe profile asks for experience with AI agents and names LangChain, LangGraph, LlamaIndex, Hugging Face or similar frameworks. Suggested preparation: explain how you would define workflow steps, inspect intermediate behavior and decide whether a framework helps or adds unnecessary complexity.

How would you expose an AI capability through a REST API for an enterprise application?

What to coverDeveloping REST APIs and integrating AI capabilities into enterprise applications are stated duties. Suggested preparation: cover request and response design, error handling, input validation and the trade-offs of a potentially slow or unreliable model dependency.

How would you deploy and monitor a containerized AI application?

What to coverThe posting calls for deployment, monitoring and maintenance in cloud and containerized environments, with familiarity in Docker, CI/CD and Azure, AWS or GCP. Suggested preparation: describe a deployment path and the health, latency, failure and quality signals you would inspect after release.

How would you test and debug a failing AI-powered application?

What to coverTesting, debugging, documentation and code reviews are explicit code-quality responsibilities. Suggested preparation: distinguish deterministic software defects from variable model outputs, and explain how tests and diagnostics would help isolate a fault.

How would you apply privacy, security and Responsible AI practices to a RAG application?

What to coverThe role requires Responsible AI, security and data governance practices, and the profile mentions data privacy. Suggested preparation: discuss controls over data access, what enters prompts and retrieved context, and how you would test for unsafe or inappropriate outputs.

Behavioural practice

Tell me about a time you debugged a difficult application issue.

What to coverSuggested preparation for the posting's debugging and maintenance duties: use a real situation, the investigation and actions you took, and the result. Explain how you narrowed down possible causes and verified the fix.

Describe a time a code review improved your team's work.

What to coverSuggested preparation for the stated code-review and code-quality responsibilities: give the situation, your contribution and the result. Explain how you handled feedback or disagreement while keeping the discussion focused on the code.

Tell me about a time you improved an AI workflow after evaluating its results.

What to coverSuggested preparation for the role's application-enhancement and RAG-optimization duties: describe a real situation, the evidence you examined, your actions and the outcome. Be specific about the trade-off you considered rather than claiming an improvement without a measure.

Describe a time you raised a privacy, security or responsible-use concern.

What to coverSuggested preparation for the posting's governance and Responsible AI responsibilities: give the situation, what you did and the result. Explain how you assessed the concern and worked toward a practical resolution.

Tell me about a time you had to learn an unfamiliar framework for a delivery task.

What to coverSuggested preparation for a role that names several AI frameworks and emphasizes professional development: use a real situation, your learning and application steps, and the result. Explain how you decided what to learn first and checked that it worked for the task.

Sharpen your knowledge

Topics to prepare

Python design and complexity
The posting calls for Python, OOP, data structures, algorithms and complexity analysis. Revision exercise: sketch two ways to deduplicate retrieved passages, compare their time and space costs, and explain which design is easier to test.
Retrieval architecture
The role explicitly covers RAG, embeddings, vector databases and semantic and hybrid search. Revision exercise: diagram a query-to-answer pipeline and explain where retrieval quality can fail and how you would compare semantic with hybrid results.
LLM prompts and agent workflows
The posting names prompt engineering, AI agents and AI-powered workflows. Revision exercise: describe a multi-step question-answering workflow, then explain how you would test its prompts and handle an irrelevant retrieved result.
LoRA and QLoRA fundamentals
The role explicitly mentions fine-tuning with LoRA and QLoRA and experience with PyTorch, TensorFlow or Keras. Revision exercise: explain how the two techniques differ and outline what you would measure when comparing a fine-tuned model with a prompt-only approach.
Responsible AI and data governance
The posting requires attention to Responsible AI, security, data privacy and data governance. Revision exercise: review a hypothetical enterprise RAG design and identify where sensitive data could enter prompts, retrieval results or logs, with a mitigation for each.

Skills to revise

Build and debug a Python REST API
The role requires REST APIs, enterprise integration, testing and debugging. Suggested practice exercise: implement a small OOP-based endpoint that accepts a question and returns a structured response; add tests for invalid input and debug one failing case.
Implement and check a RAG pipeline
Building and optimizing RAG pipelines and vector search is an explicit responsibility. Suggested practice exercise: index a small document set with embeddings, retrieve passages for several questions, and record which queries return irrelevant or missing evidence.
Prototype an AI workflow
The posting names AI agents and frameworks including LangChain, LangGraph and LlamaIndex. Suggested practice exercise: use one listed framework to connect retrieval, prompting and a response step; test what happens when retrieval returns no useful passage.
Run a small adapter fine-tuning experiment
Fine-tuning with LoRA or QLoRA is explicitly mentioned. Suggested practice exercise: using Hugging Face and PyTorch, run a small LoRA experiment on sample data and compare outputs against the unchanged model; treat the setup as practice, not a specified employer project.
Package and maintain an AI service
The role includes cloud or containerized deployment, monitoring and maintenance, and lists Docker, Git and CI/CD familiarity. Suggested practice exercise: containerize the practice API with Docker, version its tests in Git, and outline a CI/CD check and basic failure-monitoring plan without assuming a particular cloud platform.

Your preparation checklist

Prepare a GenAI project walkthrough
The role calls for LLM applications and AI-powered workflows. Prepare a concise walkthrough of a real project, if available, covering the problem, your decisions, results and what you would improve; otherwise, use a clearly labeled design example.
Sketch a RAG design
The posting emphasizes RAG, embeddings, vector databases and semantic or hybrid search. As a practice exercise, diagram ingestion, retrieval and generation, then explain how you would test retrieval quality and handle weak results.
Show prompt and model choices
Prepare a short comparison of prompt iteration, an agent workflow and LoRA or QLoRA fine-tuning: when each might help, what evidence you would seek and which listed framework you could use.
Rehearse Python and API explanations
The profile specifies Python, OOP, algorithms, complexity analysis, debugging and APIs. Prepare one code example or practice solution, plus a REST API sketch showing input validation, error handling and tests.
Map out production operations
The role includes cloud or containerized deployment, monitoring, maintenance and code quality. Prepare a deployment diagram and a brief account of how you would use Docker, Git, CI/CD, reviews and documentation.
Prepare a governance scenario
The posting calls for Responsible AI, security, data privacy and data governance. Practice explaining safeguards you would consider for sensitive data entering a RAG pipeline or appearing in model responses.
Write targeted questions for Capgemini
Ask which LLM applications and retrieval systems the team maintains, how it evaluates quality in production, and how API integration, monitoring and governance responsibilities are divided.

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…