New opportunity

Associate AI/ML Engineer

Solventum · Bangalore, India

About this role

Solventum’s Global Supply Chain Digitalization AI & Automation team seeks an Associate AI/ML Engineer to help develop and maintain production-ready machine learning, generative AI and intelligent automation solutions for supply chain operations. The role works under the guidance of senior engineers and data scientists and is based in Bangalore, India, on a hybrid arrangement; the posting specifies 40% commuting to the office.

Responsibilities

Help design, develop and deploy scalable ML models and pipelines for forecasting, optimization, predictive analytics, anomaly detection and automation.
Support the design, testing and deployment of AI agents and multi-agent workflows using LLMs, tools, memory and orchestration to improve business processes and decisions.
Develop data ingestion, transformation and feature-engineering pipelines for structured, semi-structured and unstructured enterprise data. Support integrations with knowledge repositories, knowledge graphs, vector databases and intelligent document processing.
Assist with AWS and Microsoft Azure deployments and MLOps/LLMOps processes, including versioning, CI/CD, automated deployment, monitoring and retraining. Evaluate ML models and LLMs; help improve accuracy, inference latency, cloud resource use and responsible AI practices.
Work with supply chain stakeholders, data scientists, software engineers, cloud architects, product managers and analytics teams to translate requirements into solutions. Participate in code reviews and technical documentation.

Minimum qualifications:

A bachelor’s degree in computer science, software engineering, AI or a related field with 2–4 years of professional experience in ML, AI, data science or AI engineering; alternatively, a master’s degree with relevant industry experience and 2–3 years of experience with internship experience.
At least 3 years of hands-on Python experience and foundational knowledge of ML and AI solution development, including predictive modeling, forecasting, feature engineering and model evaluation. Exposure to generative AI and agentic workflows, such as LangGraph, LangChain, AutoGen or CrewAI, through work, coursework or projects.
Foundational understanding of LLMs, prompt engineering, RAG and AI evaluation; basic understanding of system design, microservices, APIs, Docker, Kubernetes and infrastructure automation. Exposure to enterprise data platforms, SQL and distributed processing; familiarity with, or willingness to learn, AI monitoring and evaluation tools. Azure, Databricks or AWS experience is ideal.

Preferred

Academic, internship or project experience developing AI/ML solutions in supply chain, retail, healthcare or other enterprise domains. Applicants must provide education and work history by uploading a résumé or entering the details in the application.

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

Skills for this role

PythonMachine LearningGenerative AIAI AgentsMulti-agent systemsLarge Language ModelsPrompt EngineeringRAGPredictive modelingForecastingOptimizationAnomaly detectionFeature engineeringModel evaluationLangGraphLangChainAutoGenCrewAIAWSMicrosoft AzureDatabricksData ingestionData transformationData pipelinesKnowledge graphsVector databasesIntelligent document processingMLOpsLLMOpsModel versioning reactifact peiscompaigns

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

Production-ready supply chain ML
The role supports end-to-end pipelines for forecasting, optimization, predictive analytics and anomaly detection under senior guidance. Prepare to explain how a model moves from a business need through evaluation to a reliable deployment.
Agentic AI workflows
The posting calls for supporting the design, testing and deployment of LLM agents and multi-agent workflows using tools, memory and orchestration. Prepare to explain a workflow's purpose, how its components interact and how you would test its outputs.
Enterprise data integration
The role helps build ingestion, transformation and feature-engineering pipelines across structured, semi-structured and unstructured data, with integrations such as knowledge repositories and vector databases. Prepare to discuss data preparation choices and their effect on downstream results.
Cloud operations and model evaluation
The role supports AWS and Azure deployments, versioning, CI/CD, monitoring and retraining. It also helps evaluate models and LLMs for accuracy, latency, cloud resource use and responsible AI. Prepare to describe how you would measure and monitor those outcomes.
Stakeholder translation and engineering practice
The posting emphasizes working with supply chain stakeholders and technical teams to turn requirements into measurable business value, while participating in code reviews and documentation. Prepare to explain how you would clarify a use case and communicate implementation decisions.

Likely challenges to think through

Balancing quality, speed and cloud use
A likely trade-off, not a reported Solventum problem: improving model or LLM accuracy may increase inference latency or resource use. Prepare to compare candidate approaches using relevant metrics and explain when you would seek senior guidance on the trade-off.
Making mixed enterprise data usable
A likely difficulty is preserving useful context while transforming different data formats for ML or retrieval. Prepare to walk through checks for data quality and feature or retrieval relevance, and how you would detect problems before deployment.
Testing agent workflows beyond a demo
A likely difficulty is assessing workflows with LLMs, tools and memory when individual steps can produce inconsistent results. Prepare to describe representative test cases, failure checks and monitoring signals; the posting does not specify a particular testing method.

Practice your answers

Technical practice

Practice question: How would you evaluate a supply-chain forecasting model before deployment?

What to coverThe role contributes to forecasting pipelines and model evaluation. Suggested answer focus: explain a time-based validation split, a simple baseline, error metrics suited to the business decision, and checks for data leakage.

Practice question: How would you assess an anomaly-detection model when unusual events are rare?

What to coverThe posting includes anomaly detection and predictive analytics. Suggested answer focus: discuss the trade-off between missed anomalies and false alerts, how you would choose evaluation measures, and how stakeholders would review whether alerts are useful.

Practice question: How would you turn model predictions into an optimization recommendation?

What to coverOptimization and intelligent decision-making are stated areas of work. Suggested answer focus: distinguish a prediction from a recommended action, identify constraints and an objective with stakeholders, and explain how you would compare a proposed approach with the current process.

Practice question: How would you build reliable features from mixed enterprise data?

What to coverThe role supports ingestion, transformation, and feature engineering for structured, semi-structured, and unstructured data; the posting also mentions SQL and distributed processing. Suggested answer focus: describe validation, consistent feature definitions, handling missing or late data, and when processing at scale may be needed.

Practice question: How would you design a basic retrieval-augmented generation workflow for enterprise information?

What to coverThe posting calls for foundational RAG knowledge and integration with knowledge repositories, vector databases, knowledge graphs, and document processing. Suggested answer focus: trace document ingestion through retrieval and answer generation, then discuss source quality, access controls, and ways to check whether retrieved context supports the answer.

Practice question: How would you test an LLM agent that uses tools and memory?

What to coverThe role supports AI agents and multi-agent workflows involving LLMs, tools, memory, and orchestration. Suggested answer focus: explain when a simpler workflow may suffice, how you would test tool use and state handling, and how you would contain failures before deployment.

Practice question: How would you evaluate and improve an LLM-based application?

What to coverThe posting mentions prompt engineering, AI evaluation, responsible AI, accuracy, and inference latency. Suggested answer focus: propose representative test cases, assess answer quality and unsafe or unsupported outputs, and explain the trade-offs among quality, latency, and cloud resource use.

Practice question: How would you package an ML service for deployment on AWS or Azure?

What to coverThe role assists with cloud deployments and calls for a basic understanding of APIs, microservices, Docker, Kubernetes, and infrastructure automation. Suggested answer focus: outline an API and container-based deployment at a foundational level, including configuration, scaling, and operational checks without assuming a specific company architecture.

Practice question: What would you include in a production ML or LLM lifecycle?

What to coverThe posting explicitly lists versioning, CI/CD, automated deployment, monitoring, and retraining under MLOps and LLMOps. Suggested answer focus: describe how code, data or model versions remain traceable, what tests gate a release, and what monitored changes would prompt investigation or retraining.

Practice question: How would you investigate rising inference latency in a deployed AI application?

What to coverThe role supports monitoring and efforts to reduce inference latency and cloud resource use; familiarity with observability tools is listed. Suggested answer focus: separate application, model, data, and infrastructure causes using measurements, then explain how you would verify a fix without sacrificing output quality.

Behavioural practice

Practice question: Tell me about a time you translated an ambiguous stakeholder need into a technical task.

What to coverThe role works with supply-chain stakeholders to turn requirements into AI/ML solutions with measurable business value. Suggested answer focus: use a real situation, the questions and trade-offs you addressed, and the resulting agreed scope or outcome.

Practice question: Describe a time you incorporated guidance from a more experienced technical colleague.

What to coverThe posting says this associate role works under the guidance of senior engineers and data scientists. Suggested answer focus: give the situation, how you sought or applied feedback, and what changed in the work as a result.

Practice question: Tell me about a cross-functional disagreement on an AI or data solution.

What to coverThe role collaborates with data scientists, software engineers, cloud architects, product managers, and analytics teams. Suggested answer focus: describe the differing priorities, your actions to clarify evidence and trade-offs, and the eventual decision or result.

Practice question: Describe a time a code review or documentation effort improved a deliverable.

What to coverParticipating in code reviews, following engineering standards, and creating technical documentation are stated duties. Suggested answer focus: explain the original issue, your specific contribution, and the effect on clarity, reliability, or maintainability.

Practice question: Tell me about a time you investigated a problem after a model or data pipeline was deployed.

What to coverThe role helps maintain production-ready applications and supports monitoring and retraining. Suggested answer focus: use a real situation to show how you identified the issue, collaborated or escalated appropriately, and verified the outcome.

Sharpen your knowledge

Topics to prepare

Forecasting and predictive-model evaluation
The posting calls for forecasting, predictive modeling, anomaly detection and model evaluation. Suggested exercise: explain how you would choose a time-based validation split, compare a simple baseline with a model, and check for data leakage in a supply-chain forecast.
LLMs, retrieval and agent workflows
The posting asks for foundational knowledge of LLMs, prompt engineering, RAG and AI evaluation, plus exposure to agentic workflows. Suggested exercise: diagram a workflow that retrieves from an enterprise knowledge repository, passes context to an LLM and uses a tool; identify where you would test retrieval quality and answer accuracy.
Enterprise data and feature design
The role involves structured, semi-structured and unstructured data, feature engineering, SQL and exposure to distributed processing. Suggested exercise: sketch how source records and documents would be ingested and transformed into features, including checks for missing data and inconsistent identifiers.
Production architecture and MLOps
The posting mentions APIs, microservices, Docker, Kubernetes, infrastructure automation, versioning, CI/CD and retraining. Suggested exercise: draw a basic model-serving architecture and explain how a new model version would be tested, deployed, monitored and rolled back.
Performance, observability and responsible AI
The role supports monitoring ML models and LLMs, improving accuracy, inference latency and cloud resource use, and responsible AI practices under senior guidance. Suggested exercise: choose metrics for an ML prediction and an LLM response, then describe what you would investigate if quality dropped or latency rose.

Skills to revise

Build and check a Python ML pipeline
The posting specifies 3+ years of hands-on Python and exposure to feature engineering and model evaluation. Suggested exercise: write a small Python pipeline for a sample supply-chain time series, create lagged features, fit a baseline and report results on a later time period.
Query and transform enterprise data
The posting seeks exposure to SQL, data pipelines and enterprise data platforms. Suggested exercise: write SQL to aggregate transactions by item and date, then define checks for duplicates, nulls and late-arriving records before using the output as model input.
Prototype a bounded agent workflow
The role supports testing agents and multi-agent systems using frameworks such as LangGraph, LangChain, AutoGen or CrewAI. Suggested exercise: use one of those named frameworks to prototype a retrieval-and-tool workflow, and test a failed tool call and an unsupported answer; this is preparation, not a specified project.
Package and document an inference service
The posting calls for basic understanding of APIs and Docker and participation in technical documentation and code reviews. Suggested exercise: expose a sample prediction through an API, package it with Docker, and document inputs, outputs, error handling and one code-review concern.
Plan cloud deployment and monitoring
The role assists with AWS and Azure deployment and MLOps/LLMOps; the posting names monitoring tools as examples, not mandatory tools. Suggested exercise: outline a deployment on either named cloud, including version tracking, an automated deployment check, monitoring signals and a retraining trigger.

Your preparation checklist

Organize your application evidence
Prepare an accurate summary of education, work history, Python experience and relevant ML or AI work. The posting requests education and work history through a résumé upload or application fields.
Build an ML project walkthrough
Choose a real work, internship or academic example if available. Prepare a concise walkthrough of the problem, Python pipeline, feature engineering, model choice, evaluation and results; connect it to forecasting, predictive analytics or anomaly detection where applicable.
Practice an enterprise data design
As an interview exercise, sketch how you would ingest and transform structured and unstructured data, use SQL where appropriate, and produce features for a supply-chain model. Mark assumptions rather than presenting the design as a Solventum system.
Explain an agent or RAG workflow
Prepare a real example or clearly labeled practice design showing an LLM, retrieval or tools, memory or orchestration, and tests for answer quality and responsible use. The role includes supporting agent and multi-agent workflows under senior guidance.
Sketch a production deployment path
Draw a high-level design for an API-based ML or AI service on AWS or Azure. Be ready to discuss containers, versioning, CI/CD, monitoring, retraining, inference latency and cloud resource use at the foundational level described in the posting.
Prepare a collaboration example
Select an actual example, if available, of clarifying requirements with stakeholders or working with engineers and data scientists. Note your contribution, any code review or documentation, and how you checked whether the solution addressed the business need.
Write questions for the team
Prepare questions about the supply-chain problems prioritized for this associate role, how senior engineers guide design and reviews, how model and LLM performance is measured, and what the stated 40% office commuting arrangement means in practice.

Availability last checked: 4 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…