New opportunity

Data Scientist/AI-ML

DigitalXNode · Delhi, India

About this role

DigitalXNode seeks a Data Scientist/AI-ML professional to design, develop, deploy and optimize scalable AI applications for enterprise business needs. The full-time role is on-site in Delhi and specifies 1 to 5 years of experience. It combines machine learning and data science with generative AI, natural language processing and production application development.

Responsibilities

Build predictive analytics, anomaly detection, recommendation and semantic search solutions. Develop generative AI applications using large language models, LangChain workflows, intelligent agents and multi-step automation.
Implement retrieval-augmented generation systems, including document processing, chunking, indexing and context retrieval. Configure vector databases for embedding storage and search.
Develop data collection, cleansing, transformation and feature-engineering pipelines. Train, evaluate, fine-tune and optimize models; perform validation, hyperparameter tuning, benchmarking and error analysis.
Integrate models into applications through REST APIs and backend services. Deploy solutions, monitor model performance, inference latency and operational metrics, and troubleshoot pipeline, model and deployment issues. Maintain observability, telemetry and technical documentation while following security, compliance, scalability and performance practices.
Collaborate with software engineers, data teams, architects, product stakeholders and other teams; participate in Agile development and research emerging AI technologies.

Qualifications and skills:

Expertise in Python, data engineering, model training, API integration and production AI deployment. The posting identifies LangChain, PyTorch, TensorFlow, Scikit-Learn, cloud-based deployments, SQL and NoSQL databases, Git, CI/CD concepts, statistical analysis, testing and cross-functional communication as relevant skills.
A bachelor’s degree in Computer Science, Data Science, Information Technology, Statistics, Mathematics, Engineering or a related field is listed. Candidates from other educational backgrounds with strong technical expertise, project experience and demonstrated AI/ML capabilities are also encouraged to apply.
A master’s degree in an AI-, ML-, data science- or computer science-related discipline is preferred. Relevant certifications in AI, machine learning, data engineering, cloud technologies or data science are an added advantage.

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

Skills for this role

Machine LearningDeep LearningPredictive AnalyticsGenerative AILarge Language ModelsPythonObject-Oriented DesignLangChainAI AgentsWorkflow AutomationRetrieval-Augmented GenerationPrompt EngineeringNatural Language ProcessingText AnalyticsSemantic SearchPyTorchTensorFlowScikit-LearnData EngineeringFeature EngineeringData PipelinesVector DatabasesEmbeddingsSimilarity SearchREST APIsBackend IntegrationSQLNoSQLQuery OptimizationModel TrainingEvaluation and Fine-Tuning

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 AI applications for business use cases
The posting calls for predictive analytics, anomaly detection, recommendations and semantic search. Prepare to explain how a business objective becomes a measurable model or application outcome.
Build reliable data and model workflows
The role includes data collection, cleansing, transformation, feature engineering, training, validation, benchmarking and error analysis. Prepare to walk through how data quality and evaluation results guide model changes.
Implement RAG and generative AI workflows
The posting specifies LLM applications, LangChain workflows, agents, document processing, chunking, indexing, retrieval and vector databases. Prepare to explain the purpose of each stage and how to evaluate the resulting answers.
Deploy and operate integrated services
The role includes Python solutions, REST API and backend integration, deployment, monitoring of inference latency and operational metrics, and troubleshooting. Prepare an end-to-end explanation that covers both model behavior and service reliability.
Coordinate delivery and document decisions
The posting calls for collaboration with engineers, data teams, architects and product stakeholders, plus Agile participation and technical documentation. Prepare to explain how requirements, interfaces and model decisions would be communicated across those groups.

Likely challenges to think through

Balancing retrieval quality with response time
Likely preparation challenge, not a confirmed employer problem: chunking, indexing and context retrieval can improve answer relevance while increasing retrieval or inference latency. Prepare to discuss how you would compare retrieval results, answer quality and latency before changing the design.
Finding failures across the full pipeline
Likely preparation challenge, not a confirmed employer problem: a poor prediction or failed API response may originate in source data, transformations, the model or deployment. Prepare a diagnostic sequence using validation results, telemetry and operational metrics to isolate the cause.
Reconciling model performance with enterprise constraints
Likely preparation challenge, not a confirmed employer problem: improving accuracy or adding AI workflow steps may affect scalability, security or performance. Prepare to explain how you would surface those trade-offs to stakeholders and document the chosen approach.

Practice your answers

Technical practice

Suggested practice question: How would you choose an approach for predictive analytics, anomaly detection or recommendations?

What to coverCompare the available data, labels and business objective before selecting a baseline. Explain how you would measure usefulness and what trade-offs might change your choice.

Suggested practice question: How would you build a reliable data and feature-engineering pipeline?

What to coverWalk through collection, cleansing, transformation and preparation for training and inference. Explain how you would test data quality and prevent inconsistent features across those stages.

Suggested practice question: How would you validate and improve a machine-learning model?

What to coverDiscuss a suitable baseline, validation design, metrics, hyperparameter tuning and benchmarking. Explain how error analysis would guide the next change rather than relying on a single aggregate score.

Suggested practice question: How would you design a document-processing pipeline for RAG?

What to coverCover document preparation, chunking, indexing and context retrieval, all named in the posting. Discuss how chunk size and retrieved context could affect answer quality and latency.

Suggested practice question: How would you assess embeddings and vector search for semantic retrieval?

What to coverExplain embedding storage, similarity search and how you would test whether retrieved results are relevant. Consider retrieval quality, index updates and query performance as design trade-offs.

Suggested practice question: When would you use an LLM workflow, an agent or simpler automation?

What to coverThe role mentions LangChain workflows, intelligent agents and multi-step automation. Compare predictability, error handling and evaluation, and explain how prompt and context design affect the result.

Suggested practice question: How would you structure and test a Python ML application?

What to coverDescribe separations between data processing, model logic and application code, drawing on the posting’s Python and object-oriented design skills. Explain what you would test before integrating a model into another service.

Suggested practice question: How would you expose a model through a REST API?

What to coverCover request validation, inference, response design and failure handling for backend integration. Discuss how testing and documentation would help another application use the service reliably.

Suggested practice question: How would you diagnose a deployed model that becomes slower or less accurate?

What to coverUse the posting’s monitoring areas—model performance, inference latency, telemetry and operational metrics—to distinguish data, model and deployment issues. Explain how you would isolate the cause and verify a fix.

Suggested practice question: How would you balance security, scalability and performance in an AI deployment?

What to coverDiscuss how data handling, service load and inference cost could influence a cloud-based deployment design. Frame specific controls or architecture choices as proposals, since the posting does not prescribe an infrastructure.

Behavioural practice

Suggested practice question: Tell me about working across engineering, data and product teams on a technical deliverable.

What to coverUse a real situation, your actions and the result. Explain how you clarified requirements or trade-offs with people who had different priorities.

Suggested practice question: Describe a time you investigated a data, model or deployment problem.

What to coverUse a real situation, your diagnostic actions and the result. Distinguish what evidence identified the cause from what you initially suspected.

Suggested practice question: Tell me about adapting your work during an Agile sprint or technical review.

What to coverUse a real situation, what changed, your actions and the result. Explain how you communicated the effect on scope or delivery to the team.

Suggested practice question: Describe a time you evaluated a new technical approach before adopting it.

What to coverUse a real situation, your evaluation actions and the result. Discuss what evidence or trade-offs informed the decision, reflecting the role’s research and continuous-improvement duties.

Suggested practice question: Tell me about documenting or handing off a technical solution.

What to coverUse a real situation, what you documented, your actions and the result. Explain how the documentation helped others understand an architecture, API, model or training process.

Sharpen your knowledge

Topics to prepare

Model selection and evaluation
Suggested exercise: Explain how you would choose a baseline, validation approach and error analysis for predictive analytics versus anomaly detection. Posting relevance: The role explicitly includes both use cases, model validation, benchmarking and hyperparameter tuning.
RAG and semantic retrieval
Suggested exercise: Diagram how documents move through chunking, embeddings, indexing and context retrieval, then explain how you would assess poor search results. Posting relevance: The role explicitly calls for RAG systems, vector databases and semantic search.
LLM workflows and agents
Suggested exercise: Describe a multi-step LangChain workflow, including what context each step receives and how you would handle an incorrect intermediate result. Posting relevance: The role names LLM applications, LangChain workflows, intelligent agents and automation.
Data and feature pipelines
Suggested exercise: Trace a dataset from collection through cleansing, transformation and feature preparation; identify where data-quality checks belong. Posting relevance: These pipeline stages and feature engineering are stated responsibilities.
Production AI reliability
Suggested exercise: Sketch a model-serving architecture and explain which performance, latency and operational signals you would monitor, alongside security and scalability concerns. Posting relevance: The role includes deployment, observability, troubleshooting and production best practices.

Skills to revise

Python data preparation
Suggested exercise: Write a small Python pipeline that cleans records, transforms fields and produces documented model features. Posting relevance: Python development and data-processing pipelines are explicit requirements.
Model training and diagnostics
Suggested exercise: Train a baseline with Scikit-Learn, compare validation results after a small hyperparameter search, and record the main error patterns. Posting relevance: Scikit-Learn, training, tuning and error analysis appear in the posting.
RAG implementation
Suggested exercise: Build a small LangChain-based retrieval flow that chunks documents, stores embeddings in a vector database and retrieves context for a question; inspect failed retrievals. Posting relevance: The posting specifically names these RAG components and LangChain.
REST API integration
Suggested exercise: Expose a model prediction through a REST endpoint, test valid and invalid requests, and document its inputs and outputs. Posting relevance: The role calls for backend integration, RESTful APIs, testing and API specifications.
Deployment troubleshooting and monitoring
Suggested exercise: Rehearse a deployment review: check inference latency and operational metrics, diagnose a simulated model or pipeline failure, and write a short incident note. Posting relevance: Deployment, telemetry, monitoring, troubleshooting and documentation are stated responsibilities.

Your preparation checklist

Map evidence to the role
Make a one-page inventory of relevant Python, data engineering, model training, API integration and deployment work. Note which listed tools you have actually used; the posting names LangChain, PyTorch, TensorFlow and Scikit-Learn but does not require a project in every tool.
Prepare an ML project walkthrough
If you have a relevant project, outline the business problem, data collection and cleaning, feature engineering, baseline, validation, tuning, error analysis and measured results. Choose a predictive, anomaly-detection or recommendation example if available; do not invent outcomes.
Practice a RAG design exercise
Draw a clearly hypothetical document-to-answer flow covering processing, chunking, embeddings, vector indexing, context retrieval and LLM response. Explain how you would test retrieval quality and handle unsuitable or missing context; the posting explicitly calls for RAG and semantic search.
Explain application integration and operations
Prepare a small architecture diagram or API specification showing how a Python model service could connect to a backend through REST. Include proposed tests, deployment considerations, latency and model-performance monitoring, telemetry, and security safeguards; label any design choices as proposals.
Rehearse a troubleshooting example
Prepare one truthful example, or a labeled hypothetical, of diagnosing a data-pipeline failure, model error or inference bottleneck. Show the signals checked, the fix or experiment, and how you would verify the result.
Prepare a collaboration account
Outline a real instance, if available, of coordinating with engineers, data colleagues or product stakeholders: the decision needed, your contribution, communication trade-offs and documentation produced. The posting also mentions Agile planning and technical reviews.
Write targeted employer questions
Ask which use cases take priority, how RAG and conventional ML work are balanced, what deployment and monitoring stack is used, and how success is measured. Confirm expectations for the advertised full-time, on-site Delhi 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…