- SAH Diagnostics
- Bristol, West of England
- 14 days ago
Machine Learning Intern.
Before you go
Before you leave us, sign up for our email alerts
We don't do job spam, just the best digital jobs delivered straight to your inbox.
Machine Learning Intern: our view in 3 lines...
- The Role:This role is for someone building deep-learning models that classify skin lesion images as malignant or benign to support medical diagnosis.
- The Person:The person will train and fine-tune deep-learning classifiers on multi-view image sets, build a multi-modal model combining images and clinical information, and measure sensitivity, specificity, PPV and calibration.
- Requirements:The ideal candidate has experience building deep-learning classifiers in PyTorch or TensorFlow and evaluating models using held-out test sets, calibration, and subgroup analysis.
About the role
About Us
SAH Diagnostics is a growing healthcare provider dedicated to delivering high‐quality diagnostic services. We pride ourselves on compassionate patient care and professional excellence.
The role
We are building an AI skin cancer classification model. You will train and test the model on skin lesion images. Part-time hours possible.
What you will do
- Train and fine-tune deep-learning classifiers (malignant vs benign) on multi-view image sets.
- Build a multi-modal model that combines images and clinical information.
- Find which images and combinations work best (each view alone, and combined).
- Measure sensitivity, specificity, PPV and calibration.
- Report performance by skin tone.
You must have
- Experience building deep-learning classifiers (PyTorch or TensorFlow).
- Currently studying, or recently graduated, in computer science, AI, engineering or a related subject.
- Experience evaluating models properly: held-out test sets, calibration, subgroup analysis.
- The right to work in the UK. We cannot offer visa sponsorship.
It helps if you have
- A deep-learning project (PyTorch or TensorFlow) on skin lesion or medical images: dissertation, coursework or personal.

