Deep Learning Engineer – Foundation Model Training
Train transformer models on the language of tissue to build the world's most advanced foundation model for pathology.
About the Role
TileBio is building a foundation model of pathology. We treat histology as a language of tissue and train models directly on raw, unlabelled whole slide images to learn the underlying structure of biology. Our system converts tissue into a sequence of learned tokens and trains transformer models to understand disease at scale.
You will work at the core of the company. This role spans applied research and engineering. You will design and train models, build the systems that scale them, and directly shape the architecture of our foundation model. This is not a "model implementation" role. You will be expected to push methods forward, have input on architectural decisions, and own outcomes end to end.
What You'll Do
- Design and implement self supervised learning systems for histopathology at scale
- Develop representation learning methods for tile, cell, and slide level understanding
- Build and train transformer models over learned tissue tokens
- Design distributed training systems across multi GPU and multi node environments
- Run large scale experiments, analyse failure modes, and iterate rapidly
- Improve training efficiency, stability, and scalability
- Work closely with data engineering to build reproducible, high throughput pipelines
- Contribute to research direction, including new modelling approaches and training strategies
- Apply models to real world clinical and pharma problems
What We're Looking For
- Strong experience training deep learning models in PyTorch
- Hands on experience with self supervised learning such as VICReg, DINO, BYOL, or similar
- Solid understanding of representation learning and modern deep learning systems
- Experience working with transformer architectures or sequence models
- Experience with distributed training (FSDP, DDP, or equivalent)
- Strong engineering ability, not just modelling
- Ability to reason about systems, not just code
Nice to Have
- Experience working on large scale training problems
- Work that demonstrates original thinking, not just application
- Experience with large image datasets or multimodal data
- Evidence of shipping models or impactful research
What We Offer
- Salary of £45,000 to £65,000 and equity eligibility
- Work on a genuinely new paradigm in medical AI and build a foundation model from first principles
- High ownership and direct influence on company direction
- Close collaboration with leadership and domain experts
- A culture that optimises for impact, values speed, clarity, and rigour
To apply, email careers@tilebio.com with the subject "Deep Learning Engineer", your CV and a short note about why you're interested.