Lead Machine Learning Researcher

RGIT Australia; Singapore; On-site; SGD 150,000 - 210,000; Full time

Job summary

The Cambridge CARES BloodCounts! team in Singapore seeks an experienced ML researcher to lead the design, training and evaluation of multi-modal foundation models for blood analysis.

You will develop generative and self-supervised representations spanning tabular, imaging and other data modalities. Immediate start on a fixed-term two-year contract, with opportunities to influence clinical narratives and contribute to open research outputs.

Qualifications

• Substantial machine learning research experience in data-intensive settings.
• Strong track record in modern deep learning and generative modelling.
• Experience developing foundation models and adapting them to downstream tasks.
• Proficiency in Python and PyTorch with distributed/multi-GPU training.
• Experience modelling multi-modal data and clinical data challenges.
• Knowledge of reproducible ML pipelines and experiment tracking.

Responsibilities

• Lead the design, training and evaluation of multi-modal foundation models for blood analysis.
• Drive generative and self-supervised representations across modalities (tabular, imaging, etc.).
• Develop interpretable ML methods and quantify uncertainty for clinical biomarker discovery.
• Guide federated/co-developed ML efforts across the BloodCounts! consortium.
• Set technical direction for ML researchers and ensure reproducible workflows.

Skills

Python

PyTorch

Deep learning

Generative modelling

Distributed training

Multi-modal data

Research leadership

Communication skills

Git / version control

MLflow / Weights & Biases

Education

PhD in ML/AI

Postdoctoral experience

Tools

MLflow

Weights & Biases

Git

DVC

Job description

Full-time Contract Professional

Posted 11 Aug 2026

Description

Who are we?

We are Cambridge CARES, the University of Cambridge’s research centre in Singapore, established under the National Research Foundation (NRF) CREATE programme CAM.CREATE. Within Cambridge CARES, BloodCounts! is a collaborative research initiative bringing together A*STAR, Nanyang Technological University, National University of Singapore, National University Health System(NUHS) and SingHealth Hospitals, with the University of Cambridge and the company Sysmex, and many other partners in the UK, Belgium, The Gambia, Ghana, India, and the Netherlands.

BloodCounts! focusses on elevating the value of the complete blood count (known as full blood count in Singapore and the UK), the most common medical test globally, by developing new AI methodologies. In particular, we have access to population scale raw flow cytometry data that is used to generate the summary complete blood count report used by millions of healthcare workers on a daily basis. We have demonstrated that applying AI methods to this data allows for additional insights to be discovered, such as identifying markers for cancers, iron deficiency and causes of infection.

This National Research Foundation AI for Science (AI4S) supported BloodCounts! project will develop multi-modal foundation models of the blood using not only flow cytometry data but also cell imaging data. The foundation models will be developed using anonymised data from NUHS and SingHealth patients, with the aim of developing a population scale foundation model of blood that performs equitable for all ancestries in Singapore. The models developed will be applied to clinical studies for Stroke and Lung Cancer, with the aim to identify new markers that impact clinical treatment.

The BloodCounts!-AI4S project in Singapore will integrate into the wider BloodCounts! consortium, allowing for co-development of models and rapid testing of hypotheses in European, African, South Asian and East Asian populations. Researchers globally will be able to access our models and insights through a secure platform, APIs and open sourced code and models.

The BloodCounts! AI4S Team is led by Profs. Michael Roberts, Carola-Bibiane Schönlieb (Department of Applied Mathematics and Theoretical Physics (DAMTP), University of Cambridge, UK) and Lin Weisi (College of Computing & Data Science, Nanyang Technological University, Singapore). They are supported by Dr Nicholas Gleadall (University of Cambridge), Prof. Iain Bee Huat Tan (SingHealth), Prof. Hui Ji (NUS), Prof. Mickey Koh (St. George’s Hospital, London; ACTRIS, Singapore), Dr Hwee Kuan Lee (A*STAR), Prof. Parashkev Nachev (University College London Hospitals), Prof. Willem H Ouwehand (University of Cambridge), Dr Suthesh Sivapalaratnam (Barts Health, London) and Dr Chuen Wan Tan (SingHealth) alongside a wider group of collaborators.

Requirements:

• Substantial machine learning research experience in an academic, data-intensive or clinical context, typically gained over 5-10 years of postdoctoral research or industrial equivalent.
• A strong track record in modern deep learning, with particular depth in generative modelling (e.g. diffusion models, autoregressive or transformer architectures, variational and self-supervised approaches).
• Demonstrable experience developing foundation models, or other large-scale self-supervised or pre-trained models, and adapting them to downstream tasks through fine-tuning, transfer learning or prompting.
• Expert proficiency in Python and a mainstream deep learning framework (e.g. PyTorch), including distributed and multi-GPU training.
• Experience modelling multi-modal data, ideally integrating tabular, point-cloud or sequential, and imaging modalities within a unified architecture.
• A deep understanding of the challenges of clinical and real-world data modelling, including missing data, distribution shift, class imbalance, label noise and dataset bias.
• Experience with interpretability, explainability and uncertainty quantification for deep learning models, and an appreciation of their importance in a clinical setting.
• Sound understanding of machine learning engineering practice: version control (Git/GitHub or GitLab), experiment tracking (e.g. MLflow, Weights & Biases), reproducible training pipelines and data/model versioning (e.g. DVC).
• Experience deploying ML models to stakeholders or into production, and a realistic appreciation of the challenges involved in deployment.
• Experience in line management or supervision of ML researchers, with the ability to set scientific and technical direction for a team.
• Excellent verbal and written communication skills, with the ability to convey complex ML concepts to varied audiences, including clinical and non-technical stakeholders.
• Excellent organisation, prioritisation and time-management skills, and the ability to work in an agile manner in a changing research landscape.
• A commitment to keeping up to date with the rapidly changing AI research landscape.
• A positive, collaborative, problem-solving approach.

Desirable criteria are:

• Experience with federated, privacy-preserving or distributed learning frameworks (e.g. Flower, NVIDIA FLARE, OpenFL), and an understanding of its statistical challenges such as convergence under heterogeneous or biased data.
• Familiarity with physics-informed or mechanistic modelling, and with incorporating knowledge of the data-generating process into model design.
• Experience modelling haematological, flow-cytometry, imaging or other biomedical data.
• Experience with point-cloud or set-based architectures, and with representation learning for high-dimensional tabular data.
• Familiarity with fairness and equity considerations in clinical AI, particularly across diverse genetic ancestries and populations.
• A track record of securing external research funding, or a strong interest in contributing to grant proposals.
• Experience contributing to open-source model and code releases, and an understanding of best practice for reproducible, shareable research artefacts.
• Awareness of the regulatory and governance landscape for clinical AI (e.g. SaMD), and of data-governance regimes in Singapore or the UK.
• Experience working within a large, multi-site or international research consortium.

Responsibilities:

• Responsible for the delivery of all machine learning deliverables of the project, leading the design, training and evaluation of the multi-modal foundation model of blood.
• Lead the development of generative and self-supervised representations spanning the IR-FBC (tabular), Raw CBC (point-cloud) and blood smear imaging modalities.
• Lead on interpretability and uncertainty-quantification methods that underpin clinical biomarker discovery, including the stroke and cancer use cases.
• Lead on the machine learning aspects of federated co-development across the BloodCounts! consortium to ensure equitable model performance across populations.
• Set the scientific and technical direction for the team of ML researchers and oversee their day-to-day work.
• Work day-to-day with the Lead Research Software Engineer to ensure an ecosystem of tools is in place for rapid, reproducible ML research.
• Ensure that ML development across the team demonstrably follows reproducibility and best-practice standards.
• Serve as a role model and mentor to other researchers in the team.
• Lead or oversee the authoring of technical ML manuscripts, and contribute to open-source model and code releases.
• Maintain a deep understanding of the main research challenges and innovations required for the project.
• Contribute to National Research Foundation reporting documents as required, and contribute to grant proposals to secure external funding.

Please note that this post is mainly based in the CREATE Tower at NUS University Town, Singapore.

When is position available and for how long?

The position is available for an immediate start and is offered on a fixed-term contract of two years in the first instance, with the possibility of extension.

Company Overview

The Cambridge Centre for Advanced Research and Education in Singapore (CARES) was established in 2013 as the University of Cambridge’s first research centre outside the UK. It brings together researchers from the University of Cambridge, Nanyang Technological University and National University of Singapore to work on problems relevant to Singapore and the world at large. Cambridge CARES is funded by the Singapore government through its Campus for Research Excellence and Technological Enterprise (CREATE).