New opportunity

Machine Learning Associate Advisor

The Cigna Group · Hyderabad, India

About this role

The Machine Learning Associate Advisor designs, develops and deploys AI and machine learning solutions for Evernorth Health Services, a division of The Cigna Group, in Hyderabad, Telangāna, India.

Responsibilities

Design and implement AI/ML models using appropriate frameworks and libraries, and work with large language models from providers such as OpenAI and Anthropic.
Develop and optimize retrieval-augmented generation (RAG) systems. Collaborate with cross-functional teams to integrate AI capabilities into applications.
Conduct experiments, analyze results to improve model performance, follow advances in AI/ML and propose new solutions.

Required qualifications

8 to 11 years of development experience, including 5+ years working with large language models such as GPT-4 and Claude.
Strong Python proficiency and experience with AI/ML libraries such as PyTorch and TensorFlow; understanding of machine learning concepts and algorithms.
Familiarity with RAG systems; experience with graph databases such as Neo4j and knowledge graphs; experience with cloud-based ML platforms such as AWS SageMaker and Databricks; proficiency in data preprocessing and feature engineering.

Preferred qualifications

An advanced degree in Computer Science, AI or a related field; familiarity with MLOps tools and practices; experience with GPU acceleration and distributed computing; or contributions to AI/ML open-source projects or research publications.

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

Skills for this role

PythonPyTorchTensorFlowLarge Language ModelsGPT-4ClaudeRetrieval-Augmented GenerationNeo4jGraph DatabasesKnowledge GraphsMachine LearningAWS SageMakerDatabricksData PreprocessingFeature EngineeringMLOpsGPU AccelerationDistributed Computing

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

Design and deploy AI/ML solutions
The posting calls for designing, developing and deploying models, with strong Python, PyTorch or TensorFlow, and ML fundamentals. Prepare to explain technical decisions across a solution’s lifecycle.
Build and optimize LLM-based RAG
Working with provider LLMs and developing RAG systems are explicit duties. Graph databases and knowledge graphs are also required experience; prepare to explain when they help retrieval rather than assuming every RAG system needs them.
Integrate AI into applications
The role requires cross-functional collaboration to integrate AI capabilities into applications. Prepare an example that connects model choices to application needs and explains how work was coordinated across teams.
Improve performance through evidence
The posting emphasizes experiments, result analysis and model improvement, alongside data preprocessing and feature engineering. Prepare to describe how you measured an improvement and distinguished it from a change in the input data.

Likely challenges to think through

Balancing retrieval quality and system complexity
Preparation hypothesis: optimizing RAG may require trading better retrieval against added indexing, graph or pipeline complexity. Be ready to walk through how you would compare approaches using relevant evaluation results; the posting does not specify a particular architecture.
Choosing among LLM approaches
Preparation hypothesis: work with models from providers such as OpenAI and Anthropic may involve balancing output quality, latency and implementation constraints. Prepare a decision framework and an experiment you would use to test it, without assuming a mandated provider.
Moving from model results to application use
Preparation hypothesis: an experimental gain may not translate cleanly into an integrated application. Prepare to explain how you would coordinate with application teams, validate behavior after integration and address deployment constraints. The posting names cloud ML platforms but does not prescribe a deployment process.

Practice your answers

Technical practice

How would you choose an ML algorithm for a new problem?

What to coverThe posting calls for a strong understanding of ML concepts and algorithms. Suggested preparation: explain how the data, objective, evaluation metric and trade-offs between simpler and more complex models would guide your choice.

How would you implement and validate an ML model in Python?

What to coverThe role requires strong Python proficiency and experience with libraries such as PyTorch or TensorFlow. Suggested preparation: describe a reproducible training workflow, including checks on data, model outputs and evaluation results.

How would you evaluate LLMs from different providers for an application?

What to coverThe posting names work with LLMs from providers such as OpenAI and Anthropic. Suggested preparation: compare candidate models on task quality, reliability, latency and cost, using representative tests rather than assuming one model is best.

How would you design a retrieval-augmented generation system?

What to coverDeveloping and optimizing RAG systems is a stated responsibility. Suggested preparation: walk through document preparation, retrieval, context selection and answer generation, explaining how you would measure whether retrieved material supports the answer.

How would you diagnose poor answers from a RAG system?

What to coverThe role involves both RAG optimization and analysis of model performance. Suggested preparation: distinguish retrieval failures from generation failures, then describe targeted tests and the trade-offs of possible fixes.

When might a knowledge graph improve an AI retrieval approach?

What to coverThe posting requires experience with graph databases such as Neo4j and with knowledge graphs. Suggested preparation: explain when explicit relationships could help answer a question, and compare that approach with simpler retrieval on complexity and maintenance.

How would you prepare data and engineer features for an ML task?

What to coverData preprocessing and feature engineering are required skills. Suggested preparation: discuss quality checks, leakage prevention, feature choices and how you would verify that preprocessing is applied consistently during evaluation.

How would you structure an experiment to improve model performance?

What to coverThe role calls for conducting experiments and analyzing results. Suggested preparation: define a baseline and success metric, change one meaningful factor at a time where practical, and explain how you would interpret uncertain or mixed results.

How would you use a cloud ML platform to deploy and evaluate a model?

What to coverThe posting lists experience with cloud-based ML platforms such as AWS SageMaker and Databricks. Suggested preparation: outline the steps from training to deployment and monitoring, while explaining which platform choices depend on the application's needs.

What would you consider when scaling and maintaining an ML workload?

What to coverMLOps practices, GPU acceleration and distributed computing appear as preferred experience, not required duties. Suggested preparation: explain when reproducibility, monitoring or additional compute would help, and when their operational complexity would outweigh the benefit.

Behavioural practice

Tell me about a time you integrated an AI capability with another team's application.

What to coverCross-functional integration is a stated responsibility. Suggested preparation: use situation, action and result to show how you agreed on requirements, resolved an interface or delivery issue, and assessed the outcome.

Describe an experiment whose results challenged your initial approach.

What to coverThe role calls for experimenting and using results to improve performance. Suggested preparation: give a real situation, the analysis and action you took, and the resulting decision or measurable outcome.

Tell me about a time you proposed a new AI or ML approach.

What to coverThe posting asks the role to follow advances in AI/ML and propose solutions. Suggested preparation: explain the situation, how you evaluated the idea against the existing approach, what you proposed, and the result.

Describe a time you carried an ML solution from design toward deployment.

What to coverDesigning, developing and deploying solutions are central to the advertised role. Suggested preparation: choose a real example and describe the situation, your actions and trade-offs, and what happened after delivery.

Sharpen your knowledge

Topics to prepare

Machine learning fundamentals
Revise how to select algorithms, recognize overfitting and choose evaluation measures; explain the trade-offs for a hypothetical model. The posting requires a strong understanding of ML concepts and algorithms.
Large language model behavior
Compare when a model such as GPT-4 or Claude may need retrieved context, and explain how you would assess answer quality. The role works with LLMs from providers including OpenAI and Anthropic.
RAG architecture
Draw a retrieval-to-generation flow and explain where irrelevant retrieval could degrade an answer. Developing and optimizing RAG systems is an explicit responsibility.
Knowledge graphs and graph databases
Sketch entities and relationships for a small example, then explain when traversing those relationships would help retrieval. The posting requires experience with graph databases such as Neo4j and knowledge graphs.
Experimental design
Outline a comparison of two model or retrieval variants, including a baseline, measures and how you would interpret mixed results. The role calls for experiments and analysis to improve model performance.
Cloud ML workflow
Map the steps from prepared data to a deployed model and identify where AWS SageMaker or Databricks could fit. Experience with cloud-based ML platforms is required; revisiting MLOps practices is an additional suggestion because they are preferred, not required.

Skills to revise

Python model implementation
Implement a small training and evaluation workflow in Python using PyTorch or TensorFlow, and be ready to explain design choices. Strong Python proficiency and experience with these AI/ML libraries are required.
LLM application development
Build a small Python example that sends a task and supplied context to an LLM, then inspect several outputs for failures. This is a preparation exercise for the stated work with LLMs, not a specified employer project.
RAG prototyping
Build a small retrieval-and-answering prototype over sample documents; test cases where the relevant passage is present, absent or poorly ranked. The role explicitly develops and optimizes RAG systems.
Graph data modeling
Create a small graph in Neo4j and query relationships that answer a concrete question. This practices the required graph-database and knowledge-graph experience.
Data preparation and feature engineering
Clean a small dataset, document missing-value decisions and derive a feature; check that the feature would be available when the model is used. Both preprocessing and feature engineering are explicit requirements.
Performance comparison and integration
Run a repeatable comparison of two model or RAG variants, summarize the results and describe how the better option could be integrated into an application. This exercises the posting’s responsibilities to analyze experiments and collaborate on AI integration; the particular exercise is a suggestion.

Your preparation checklist

Build a requirements evidence sheet
Map the posting’s 8–11 years of development experience, 5+ years with LLMs, and other required skills to specific examples you can substantiate. Mark gaps rather than inventing evidence.
Prepare an AI/ML project walkthrough
Choose a real project, if available, and make a one-page account of the problem, your decisions, Python and ML tools used, deployment approach, and measured result. The role calls for designing, developing, and deploying models.
Sketch a RAG system
Prepare a hypothetical design covering document preparation, retrieval, an LLM response, and ways to improve retrieval quality. Explain where a knowledge graph or graph database such as Neo4j would help—and where it would not.
Practice a Python model exercise
Work through a small example showing preprocessing, feature engineering, model selection, and evaluation. Be ready to explain relevant PyTorch or TensorFlow choices and the underlying ML concepts.
Show how you evaluate LLM changes
Draft an experiment plan comparing two prompts, models, or RAG configurations, with a baseline, evaluation criteria, result analysis, and a decision rule. This reflects the posting’s emphasis on experiments and performance improvement.
Rehearse an integration and deployment design
Diagram how an AI capability could enter an application, including interfaces, cross-functional handoffs, and deployment considerations. Explain any relevant experience with a cloud ML platform such as AWS SageMaker or Databricks.
Prepare optional depth evidence
If applicable, select one concise example of MLOps, GPU or distributed computing, open-source work, research, or advanced study. The posting lists these as preferred, not required.
Write questions for the employer
Ask which AI applications this Evernorth team supports, how it measures model and RAG performance, and how engineering teams share integration responsibilities. Treat the answers as information to learn, not assumptions about the role.

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…