New opportunity

Machine Learning Engineer

Julius Baer · Singapore

About this role

Julius Baer seeks a Machine Learning Engineer for an agile development team in its ML & AI ART in Singapore. The role covers the lifecycle of scalable AI solutions, from concept and development through deployment and maintenance in a DevOps environment, addressing business challenges in wealth management.

Responsibilities

Develop, deploy and optimize machine learning and AI solutions. Build multi-agent systems, give AI models function- and tool-calling capabilities, and optimize agents and agentic workflows for production.
Design retrieval-augmented generation (RAG) systems that ground AI responses in enterprise data. Assess and integrate models, including LLMs, for performance and reliability; test and refine system prompts and few-shot examples for accurate, consistent and safe outputs.
Evaluate models using appropriate metrics, evaluation sets and techniques, and improve them iteratively. Work with platform teams, data engineers, data scientists and other stakeholders to integrate solutions into existing systems. Participate in code reviews, testing and debugging.
Follow applicable laws and regulations and demonstrate appropriate standards of integrity, fair dealing, conflict-of-interest management and risk management.

Requirements

A bachelor's or master's degree in Data Science, Computer Science, Mathematics, Statistics or a related field, plus relevant work experience in machine learning, data science or a related field. No minimum number of years is specified.
Strong Python programming skills and experience with machine learning libraries and deep learning frameworks such as TensorFlow or PyTorch. Proven LLM experience and an understanding of AI agents, agentic workflows, LLM orchestration frameworks and reasoning patterns.
Experience with data preprocessing, feature engineering, model selection and evaluation; knowledge of probability, linear algebra and optimization; and familiarity with version control, testing and documentation.
Strong analytical, problem-solving, debugging, communication and collaboration skills, with the ability to prioritize work, meet timelines and produce reliable, maintainable code.

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

Skills for this role

PythonMachine LearningDeep LearningTensorFlowPyTorchLarge Language ModelsAI AgentsMulti-agent SystemsAgentic WorkflowsFunction CallingRetrieval-Augmented GenerationPrompt EngineeringFew-shot PromptingLLM OrchestrationData PreprocessingFeature EngineeringModel SelectionModel EvaluationProbabilityLinear AlgebraOptimizationVersion ControlSoftware TestingDebuggingDevOpsCode ReviewRisk ManagementStakeholder Collaboration

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 agentic AI
Role: Build multi-agent systems, add function and tool calling, and optimize agentic workflows for production. Preparation suggestion: Be ready to explain design choices, testing and failure handling for an agent workflow.
Enterprise-grounded responses
Role: Design and implement RAG systems that ground AI responses in enterprise data. Preparation suggestion: Explain how you would assess retrieval quality and whether a generated answer is supported by retrieved content.
Measured model improvement
Role: Assess and integrate models for performance and reliability, evaluate them with metrics and evaluation sets, and refine prompts and few-shot examples for accurate, consistent and safe outputs. Preparation suggestion: Show how evaluation results would guide an iteration rather than relying on isolated examples.
End-to-end engineering and integration
Role: Develop, deploy and maintain scalable AI solutions in a DevOps environment; collaborate with platform teams, data engineers, data scientists and other stakeholders; and participate in code review, testing and debugging. Preparation suggestion: Be ready to explain how you keep an integrated solution reliable and maintainable after deployment.
Risk-conscious delivery
Role: The posting calls for compliance with applicable laws and regulations, adequate risk management, integrity, fair dealing and management of conflicts of interest. Preparation suggestion: Describe how you would surface and address a potential risk while developing or deploying an AI solution, without assuming a particular company control process.

Likely challenges to think through

Agent flexibility versus controlled outputs
Preparation hypothesis, not a confirmed employer problem: Tool-calling and multi-agent workflows may create more ways for outputs to vary, while the role calls for accurate, consistent and safe results. Prepare to explain how you would test agent decisions and tool use, identify failures and refine the workflow.
Retrieval coverage versus answer grounding
Preparation hypothesis, not a confirmed employer problem: A RAG system may need to answer useful questions even when retrieved enterprise content is incomplete or weakly relevant. Prepare to explain how you would evaluate retrieval and responses, and how you would handle insufficient supporting content.
Delivery pace versus maintainability
Preparation hypothesis, not a confirmed employer problem: Iterating on models while integrating with existing systems across teams may put timelines in tension with testing and reliable code. Prepare an example of how you would prioritize work, coordinate dependencies and use reviews or debugging to protect quality.

Practice your answers

Technical practice

How would you preprocess data, engineer features and select a model for a machine learning problem?

What to coverThe posting names preprocessing, feature engineering, model selection and evaluation. Suggested preparation: explain how you would check data quality and leakage, establish a baseline, and use an evaluation approach suited to the problem.

How would you use Python and a deep learning framework to build maintainable model code?

What to coverThe role calls for strong Python skills and experience with frameworks such as TensorFlow or PyTorch. Suggested preparation: discuss code structure, reproducible experiments and testing, including where probability, linear algebra or optimization inform implementation choices.

How would you design a multi-agent workflow, and when would you avoid one?

What to coverThe role explicitly includes multi-agent systems, agentic workflows, orchestration and reasoning patterns. Suggested preparation: explain how you would divide responsibilities, track intermediate results and weigh added coordination and failure modes against a simpler design.

How would you give an AI model function- or tool-calling capabilities reliably?

What to coverThe posting explicitly calls for function- and tool-calling capabilities and production-ready agents. Suggested preparation: describe tool selection, input and output validation, error handling and how you would test calls before production use.

How would you design a RAG system that grounds responses in enterprise data?

What to coverThe role explicitly requires designing and implementing RAG systems. Suggested preparation: walk through retrieval, use of retrieved context in generation, and ways to evaluate whether an answer is supported by the source material.

How would you compare LLMs for integration into an AI solution?

What to coverThe posting calls for assessing and integrating AI models for performance and reliability. Suggested preparation: explain how representative tasks and evaluation sets would inform trade-offs among output quality, reliability and production constraints.

How would you test and refine system prompts and few-shot examples?

What to coverThe role specifies accurate, consistent and safe AI outputs as goals for prompt optimization. Suggested preparation: describe controlled comparisons, representative and challenging inputs, and checks for regressions rather than relying on a few convincing examples.

How would you evaluate an ML or AI solution and decide what to improve next?

What to coverThe posting explicitly calls for appropriate metrics, evaluation sets and iterative improvement. Suggested preparation: distinguish task-level performance from reliability, explain how you would choose metrics, and use error analysis to prioritize changes.

How would you take an AI solution from concept through deployment and maintenance in a DevOps environment?

What to coverThe role spans the solution lifecycle and emphasizes robust, scalable production systems. Suggested preparation: discuss validation before release, observing performance after deployment, and the trade-off between iteration speed and reliability.

How would you diagnose a reliability issue in an integrated ML or AI solution?

What to coverThe role includes integration with existing systems, code reviews, testing and debugging alongside platform and data teams. Suggested preparation: explain how you would isolate whether data, model, agent workflow or integration is responsible, then test and document the fix.

Behavioural practice

Tell me about a time you worked with different teams to integrate a technical solution.

What to coverThe posting emphasizes collaboration with platform teams, data engineers, data scientists and other stakeholders. For practice, use a real situation, your actions to resolve differing needs, and the result of the integration work.

Tell me about a time you prioritized competing work while maintaining quality.

What to coverThe role calls for managing workloads, timelines and reliable, maintainable code. For practice, describe the situation and constraints, how you chose and communicated priorities, and the result.

Tell me about a difficult bug or reliability problem you investigated.

What to coverTesting, debugging and problem-solving are explicit parts of the role. For practice, give a real situation, the evidence and actions that led you to a diagnosis, and the result, including what you verified.

Tell me about a time you raised or managed a risk in technical work.

What to coverThe posting names integrity, due care, fair dealing, conflicts of interest and risk management. For practice, explain the situation, the action you took to surface or address the concern, and the result without implying that a particular company procedure applied.

Tell me about a time you learned a new ML or AI approach and judged whether to use it.

What to coverThe role asks for continuous learning alongside delivery of reliable solutions. For practice, describe the situation, how you assessed the approach against the task, and the result of your decision.

Sharpen your knowledge

Topics to prepare

RAG and enterprise grounding
The posting calls for RAG systems that ground AI responses in enterprise data. Revision exercise (suggestion): sketch how a response would use retrieved material, then identify where unsupported claims could enter.
Agents, orchestration and reasoning patterns
The role includes multi-agent systems, function/tool calling and production agentic workflows. Revision exercise (suggestion): diagram an agent workflow and explain when it should call a function, hand off work or stop.
Model evaluation and reliability
The posting requires appropriate metrics, evaluation sets and iterative improvement of ML and AI models. Revision exercise (suggestion): define an evaluation set and metrics for a hypothetical grounded-answer task, including how to inspect failures.
ML foundations and model selection
The requirements name preprocessing, feature engineering, model selection, probability, linear algebra and optimization. Revision exercise (suggestion): explain how data preparation and an optimization objective affect the choice and evaluation of a model.
Production lifecycle and risk
The role spans development, deployment and maintenance in a DevOps environment, alongside risk management and compliance responsibilities. Revision exercise (suggestion): trace a hypothetical model change from review through deployment and identify reliability and customer-fairness questions to raise.

Skills to revise

Python model implementation
Strong Python skills and experience with ML libraries and TensorFlow or PyTorch are requested. Revision exercise (suggestion): implement and debug a small training-and-evaluation workflow in Python using a framework you know.
Data preparation and evaluation
The posting asks for preprocessing, feature engineering, model selection and evaluation experience. Revision exercise (suggestion): prepare a small sample dataset, compare two model choices and document the metric and errors that guide your choice.
RAG workflow design
Designing and implementing RAG over enterprise data is an explicit responsibility. Revision exercise (suggestion): build a small prototype over sample documents and check whether each answer is supported by retrieved content.
Tool-calling and agent debugging
The role calls for function/tool-calling capabilities and optimized agentic workflows. Revision exercise (suggestion): implement a small tool-calling flow with a simulated function, then test incorrect inputs, failed calls and stopping behavior.
Prompt testing
The posting specifically mentions testing system prompts and few-shot examples for accurate, consistent and safe outputs. Revision exercise (suggestion): compare two prompt versions against the same evaluation cases and record where outputs improve or regress.
Maintainable delivery and collaboration
Code reviews, testing, debugging, version control, documentation and cross-functional integration are stated expectations. Revision exercise (suggestion): review a small ML code change, add a relevant test, document an integration assumption and explain the change to a platform or data-engineering stakeholder.

Your preparation checklist

Prepare an end-to-end ML project walkthrough
Choose a relevant project you can discuss and make a one-page account of the business problem, Python implementation, preprocessing, feature engineering, model selection, deployment and maintenance. Be ready to explain the mathematical reasoning behind key choices.
Sketch a grounded RAG design
As a preparation exercise, draw a system that retrieves from enterprise data and grounds an LLM response. Note how you would assess retrieval quality, response accuracy and reliability; do not assume Julius Baer uses a particular architecture.
Map an agentic workflow
Prepare a diagram showing when an agent calls a function or tool, how multiple agents would coordinate if needed, and how you would test and optimize the workflow for production use.
Build an evaluation example
Bring a small example evaluation set, chosen metrics and an iteration log showing how model choice, system prompts or few-shot examples affected accuracy, consistency and safety.
Show engineering-quality evidence
Identify code or a project you can discuss to illustrate version control, tests, documentation, debugging and code review. Explain how you handled deployment and maintenance in a DevOps setting, if applicable.
Prepare a collaboration and risk example
Outline a real example of working across teams, prioritizing work and resolving a problem. Be ready to explain how you considered integrity, fair dealing, conflicts of interest and risk management without claiming knowledge of the bank’s internal controls.
Write questions for the interviewers
Ask how the ML & AI ART works with platform teams and data specialists, how production AI solutions are evaluated and maintained, and which business challenges this role would address first.

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 Singapore. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…