Company Description
UHN is Canada’s #1 hospital and the world’s #1 publicly funded hospital. With 10 sites and more than 44,000 TeamUHN members, UHN consists of Toronto General Hospital, Toronto Western Hospital, Princess Margaret Cancer Centre, Toronto Rehabilitation Institute, The Michener Institute of Education and West Park Healthcare Centre. As Canada's top research hospital, the scope of biomedical research and complexity of cases at UHN have made it a national and international source for discovery, education and patient care. UHN has the largest hospital-based research program in Canada, with major research in neurosciences, cardiology, transplantation, oncology, surgical innovation, infectious diseases, genomic medicine and rehabilitation medicine. UHN is a research hospital affiliated with the University of Toronto.
UHN’s vision is to build A Healthier World and it’s only because of the talented and dedicated people who work here that we are continually bringing that vision closer to reality.
www.uhn.ca
Job Description
Union: Non-Union
Number of Vacancies: 1
New or Replacement Position: New
Site: MaRS
Department: PM Research
Reports to: Senior Scientist
Salary Range: $54,902 - $93,333 Per Year
Hours: 37.5 Hours Per Week
Shifts: Monday - Friday; Day Shifts
Status: Temporary Full-time (2-Year Contract)
Closing Date: May 26, 2026
Position Summary:
We are seeking a postdoctoral researcher with strong expertise in AI/ML to join a major interdisciplinary initiative focused on developing foundation models and secure computational infrastructure for translational cancer research.
The first major objective is to build a general-purpose drug foundation model capable of transfer learning across diverse prediction tasks, including mechanism of action classification, clinical drug response prediction in tumour subtypes, ADMET and toxicity profiling, combinatorial drug synergy, and drug repurposing. The goal is to move beyond task-specific architectures and datasets toward flexible models that can generalize across therapeutic contexts.
The second major objective is to develop and apply secure, scalable, and privacy-preserving computational infrastructure to support biomarker discovery across diverse treatment modalities. This will include building agentic AI approaches to harmonize clinical, genomic, and transcriptomic data across public and private cohorts, while assessing the predictive value of DNA and RNA signatures in federated settings where sensitive data remain under local governance.
Together, these efforts aim to advance AI-driven drug discovery, biomarker development, and clinical translation by integrating modern machine learning, multimodal biomedical data, and robust distributed analysis frameworks. The successful candidate will work in the Haibe-Kains Lab at the Princess Margaret Cancer Centre, University Health Network.
Duties:
- Design and implement multimodal drug foundation models that integrate molecular graph representations, bulk transcriptomic perturbation signatures, and multi-omics cell-state representations.
- Develop transfer learning strategies to support diverse drug prediction tasks, including mechanism of action classification, clinical drug response prediction, ADMET and toxicity profiling, combinatorial drug synergy, and drug repurposing.
- Build flexible AI/ML workflows that reduce reliance on task-specific architectures and enable generalization across therapeutic contexts, tumour subtypes, and treatment modalities.
- Architect and deploy secure, scalable, and privacy-preserving computational infrastructure for biomarker discovery and translational cancer research.
- Develop agentic AI-enabled frameworks to support the harmonization, annotation, quality control, and integration of clinical, genomic, and transcriptomic data across public cohorts and private institutional datasets.
- Implement distributed and federated analysis pipelines in which each contributing dataset can be analysed separately, enabling multi-cohort biomarker assessment without raw data centralization.
- Develop systematic workflows to evaluate published and user-specified DNA and RNA signatures, including immune, stromal, mutation-based, pathway-level, and treatment-response signatures.
- Assess the predictive value of molecular signatures across cancer types and treatment modalities, including chemotherapy, targeted therapy, immunotherapy, and emerging therapeutic approaches.
- Integrate biomarker discovery and immunotherapy inference pipelines with clinical data warehouses to support translational studies in collaboration with clinical, industry, and computational partners.
- Contribute to responsible data sharing frameworks, data governance processes, IRB/ethics protocols, and regulatory documentation as required.
- Collaborate closely with computational biologists, software developers, clinicians, and wet-lab scientists to build, validate, and translate predictive models and biomarker discovery workflows.
Qualifications
- Awarded a PhD within the previous 5 years, or an MD or DDS within the previous 10 years in a relevant quantitative or biomedical discipline, including but not limited to: Computational Biology, Bioinformatics, Systems Biology, or Quantitative Genomics; Machine Learning, Artificial Intelligence, Computer Science, or Data Science; Biostatistics, Biomedical Engineering, or related fields
- Demonstrated experience developing or applying deep learning methods to molecular, biological, clinical, or multi-omics data.
- Expertise in one or more modern AI/ML approaches relevant to foundation models or representation learning, such as graph neural networks, transformers, generative models, self-supervised learning, few-shot or zero-shot learning, or transfer learning.
- Strong programming skills in Python and/or R, with practical experience using modern machine learning frameworks and tooling such as PyTorch, Hugging Face, PyTorch Geometric, Deep Graph Library, scikit-learn, or equivalent platforms.
- Experience working with large-scale biomedical datasets, such as molecular graphs, transcriptomic perturbation profiles, multi-omics data, clinical genomics, electronic health record-derived data, or treatment-response datasets.
- Proficiency with reproducible workflow management systems such as Snakemake, Nextflow, CWL, or equivalent pipeline frameworks.
- Familiarity with cloud or high-performance computing environments, such as GCP, AWS, SLURM-based clusters, or equivalent infrastructure.
- Understanding of data harmonization, privacy-preserving analysis, federated learning, secure distributed computing, or clinical data governance is highly desirable.
- Strong publication record, commensurate with career stage, in computational biology, AI/ML, bioinformatics, biostatistics, biomedical data science, or related fields.
- Excellent communication skills and ability to work collaboratively in interdisciplinary teams spanning computational biology, machine learning, software engineering, oncology, and clinical research.
- Experience with clinical data harmonization (e.g., OMOP CDM, HL7 FHIR) is preferred.
- Experience designing scalable bioinformatics pipelines for large-scale genomic or transcriptomic datasets, preferred.
- Understanding of regulatory and data governance requirements in clinical research settings, preferred.
- Prior collaborative work across computational and clinical or wet-lab research teams, preferred.
Additional Information
Why join UHN?
In addition to working alongside some of the most talented and inspiring healthcare professionals in the world, UHN offers a wide range of benefits, programs and perks. It is the comprehensiveness of these offerings that makes it a differentiating factor, allowing you to find value where it matters most to you, now and throughout your career at UHN.
- Competitive offer packages
- Government organization and a member of the Healthcare of Ontario Pension Plan (HOOPP https://hoopp.com/)
- Close access to Transit and UHN shuttle service
- A flexible work environment
- Opportunities for development and promotions within a large organization
- Additional perks (multiple corporate discounts including: travel, restaurants, parking, phone plans, auto insurance discounts, on-site gyms, etc.)
Current UHN employees must have successfully completed their probationary period, have a good employee record along with satisfactory attendance in accordance with UHN's attendance management program, to be eligible for consideration.
All applications must be submitted before the posting close date.
UHN uses email to communicate with selected candidates. Please ensure you check your email regularly.
Please be advised that a Criminal Record Check may be required of the successful candidate. Should it be determined that any information provided by a candidate be misleading, inaccurate or incorrect, UHN reserves the right to discontinue with the consideration of their application.
UHN is an equal opportunity employer committed to an inclusive recruitment process and workplace. Requests for accommodation can be made at any stage of the recruitment process. Applicants need to make their requirements known.
We thank all applicants for their interest, however, only those selected for further consideration will be contacted.
