New opportunity

Senior Manager / Manager, LLM Architect, SIMFONI

Consortium for Clinical Research and Innovation Singapore (CRIS) · Singapore

About this role

The Consortium for Clinical Research and Innovation Singapore (CRIS) is seeking a Senior Manager/Manager, LLM Architect for the SIngapore Medical FOundation AI model (SIMFONI) programme. This role provides technical leadership for a healthcare foundation model initiative focused on developing, adapting, evaluating, and deploying large language models for clinical decision support and healthcare knowledge assistance. This is a technical leadership and advisory role rather than a day-to-day coding position. Responsibilities include: LLM Technical Leadership and Oversight: Serve as the programme-level technical lead for LLM architecture, development strategy, and engineering feasibility. Review and advise project teams on model selection, pretraining, fine-tuning, and evaluation design. Establish engineering guidelines and review principles. Support for Training and Evaluation: Advise on adaptation methods including RAG, grounding, and preference optimization (RLHF, DPO, GRPO). Assess evaluation strategies for clinical relevance, safety, and hallucination risk. Engineering and LLMOps Advisory: Guide teams on experiment tracking, model serving, monitoring, and continuous iteration. Identify cross-cutting engineering risks related to scalability, reliability, and cost. Cross-Team Alignment: Facilitate technical design reviews and translate business objectives into actionable technical criteria. Required Qualifications: 5+ years of experience in AI/LLM, machine learning, or large-scale technology architecture. Proven experience with LLM/foundation model development, adaptation, or deployment. Strong understanding of modern LLM architectures, training methods, and inference patterns. Ability to provide actionable technical guidance and challenge assumptions. Strong written and verbal communication skills. Preferred Qualifications: Industry experience in production-level LLM development. Experience in healthcare AI, biomedical NLP, or clinical decision support. Familiarity with healthcare data governance, privacy, and security. Experience working with multi-partner or cross-institutional programmes. Familiarity with Singapore's healthcare ecosystem is an advantage. This is a 3-year renewable contract position.

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

Skills for this role

Large Language ModelsFoundation ModelsMachine LearningContinual PretrainingFine-tuningInstruction TuningRLHFDPOGRPORAGAgentic LLM SystemsLLM EvaluationSafety TestingHallucination MitigationDistributed TrainingGPU InfrastructureModel ServingMLOpsLLMOpsModel Lifecycle ManagementTechnical LeadershipArchitecture Review

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

Technical Oversight and Architecture Governance
The role requires establishing and enforcing engineering guidelines, review principles, and technical design criteria across multiple project teams. You must be able to evaluate the feasibility, scalability, and reproducibility of proposed LLM development strategies, ensuring they align with the programme's clinical and research objectives.
Strategic Advisory on LLM Adaptation and Evaluation
You will guide teams on complex technical decisions including continual pretraining, fine-tuning, and preference optimization (RLHF, DPO, GRPO). A critical priority is defining robust evaluation frameworks that specifically address clinical relevance, safety, and hallucination mitigation in a healthcare context.
Cross-Team Technical Alignment and Risk Management
As a central technical lead, you must facilitate communication between research, engineering, and infrastructure teams. You are responsible for identifying cross-cutting risks related to model lifecycle management, cost, and operational feasibility, ensuring that disparate workstreams remain synchronized.

Likely challenges to think through

Balancing Technical Rigor with Non-Coding Leadership
The challenge lies in providing credible, actionable guidance without being the primary developer. Prepare by demonstrating your ability to review technical documentation and code-level decisions to challenge assumptions effectively, while maintaining a high-level architectural perspective that supports rather than micromanages teams.
Navigating High-Stakes Healthcare Constraints
Applying LLMs to clinical decision support introduces stringent requirements for safety, privacy, and data governance. Prepare by articulating how you would implement 'responsible AI' practices and technical review gates that satisfy both clinical safety standards and the need for rapid innovation.
Managing Multi-Partner Technical Complexity
Working across multiple institutional partners requires navigating different technical maturity levels and priorities. Prepare by highlighting your experience in building consensus among diverse stakeholders and creating standardized documentation or decision records that ensure consistency across a large-scale, multi-vendor programme.

Practice your answers

Technical practice

Evaluating LLM Adaptation Strategies for Clinical Use

What to coverDiscuss the trade-offs between continual pretraining and supervised fine-tuning when adapting a foundation model for clinical decision support. Address how you would determine which method is most appropriate given specific data availability and the need for domain-specific knowledge versus instruction following.

Mitigating Hallucination in Healthcare AI

What to coverExplain your approach to assessing and mitigating hallucination risks in LLMs deployed for clinical assistance. Include how you would design evaluation strategies that prioritize clinical safety and factual grounding over general performance metrics.

RAG vs. Fine-tuning for Medical Knowledge

What to coverWhen building a healthcare knowledge assistant, how do you decide between Retrieval-Augmented Generation (RAG) and model fine-tuning? Discuss the implications for data freshness, source attribution, and the ability to handle sensitive clinical data.

Preference Optimization Techniques

What to coverCompare RLHF, DPO, and GRPO in the context of aligning models for healthcare applications. Explain how you would advise a team on selecting an optimization method based on their available feedback data and computational resources.

LLMOps and Model Lifecycle Management

What to coverDescribe the key components of an LLMOps pipeline required to support continuous iteration of clinical models. Focus on how you would ensure reproducibility, experiment tracking, and safe deployment gates in a high-stakes environment.

Scalability and Infrastructure Considerations

What to coverWhat are the primary engineering risks when scaling LLM inference for a national healthcare programme? Address considerations regarding GPU infrastructure, latency, cost management, and model serving patterns.

Technical Governance and Review Principles

What to coverAs a technical lead, how do you establish review principles for LLM development across multiple project teams? Explain how you would define 'model readiness' criteria to ensure consistency and quality across diverse workstreams.

Data Privacy and Security in Healthcare AI

What to coverGiven the sensitivity of healthcare data, what architectural patterns or security practices would you advocate for when training or deploying LLMs? Discuss the balance between model performance and strict adherence to data governance requirements.

Behavioural practice

Navigating Technical Disagreements

What to coverDescribe a situation where you had to challenge a team's technical assumptions regarding an AI project. How did you communicate your concerns, and what was the outcome in terms of the project's direction or quality?

Translating Business Objectives to Technical Criteria

What to coverProvide an example of how you translated high-level clinical or research objectives into specific, actionable technical requirements for an engineering team. How did you ensure the team understood the alignment between their work and the end-user needs?

Managing Cross-Institutional Collaboration

What to coverShare an experience where you had to provide technical leadership across multiple partner organizations or teams with different priorities. How did you facilitate alignment and ensure that the technical design remained cohesive?

Handling Technical Risk and Uncertainty

What to coverDescribe a time you identified a significant cross-cutting engineering risk in a large-scale AI programme. How did you document, communicate, and mitigate this risk while maintaining project momentum?

Sharpen your knowledge

Topics to prepare

Clinical Safety and Hallucination Mitigation
The role requires assessing evaluation strategies for clinical relevance and safety. Exercise: Draft a technical framework for evaluating a medical LLM, specifically defining metrics for hallucination detection and clinical safety thresholds. This is relevant because the SIMFONI programme prioritizes the safe use of AI in public healthcare.
LLM Adaptation and Alignment Strategies
The position involves advising on continual pretraining, fine-tuning, and preference optimization (RLHF, DPO, GRPO). Exercise: Compare the trade-offs between domain-adaptive pretraining and RAG for a clinical decision support system. This is relevant as you will be responsible for guiding project teams on the most effective adaptation methods for healthcare knowledge assistance.

Skills to revise

Technical Architecture Review
You must provide oversight and challenge assumptions across multiple project teams. Exercise: Create a template for a technical design review document that includes sections for scalability, cost, and engineering risk. This is relevant because the role is a technical leadership position focused on establishing review principles and engineering guidelines.
LLMOps and Infrastructure Advisory
The role requires guiding teams on model serving, monitoring, and lifecycle management. Exercise: Outline a strategy for model versioning and experiment tracking in a multi-partner environment. This is relevant as you will be identifying cross-cutting engineering risks related to reliability and operational feasibility.

Your preparation checklist

Develop a framework for LLM technical design reviews
Create a template for evaluating LLM project proposals that covers model selection, training methodology (e.g., SFT, RLHF, DPO), and infrastructure requirements. This demonstrates your ability to establish the engineering guidelines and review principles required by the SIMFONI programme.
Prepare a case study on clinical LLM risk mitigation
Draft a brief technical document outlining how you would approach hallucination mitigation, safety testing, and bias evaluation in a high-stakes healthcare environment. Focus on balancing model performance with clinical safety and regulatory compliance.
Outline an LLMOps strategy for multi-team environments
Design a high-level architecture for model lifecycle management, including experiment tracking, serving, and monitoring. Explain how you would ensure reproducibility and scalability across diverse project teams, which is central to the advisory nature of this role.
Formulate strategic questions for the interview panel
Prepare questions regarding the current state of the SIMFONI infrastructure, the primary technical bottlenecks faced by existing project teams, and the specific governance frameworks used to manage healthcare data privacy and security in their AI initiatives.
Synthesize experience in cross-institutional collaboration
Prepare concrete examples of how you have previously translated complex technical requirements into actionable guidance for non-technical stakeholders or cross-functional teams, emphasizing your ability to facilitate alignment in multi-partner programmes.

Availability last checked: 18 Oct 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…