Data Scientist (Graph & Clinical Data)
Healthcare information is relational by nature. Patients, conditions, treatments and time do not sit comfortably in rows and columns, and most of what matters clinically lives in the relationships rather than the records. This role sits with that half of the problem.
You will
- Design representations of complex, time-dependent clinical information, and defend why one representation beats another
- Work with standardised medical terminologies and the hierarchies they impose
- Apply graph methods (traversal, connectivity, similarity, clustering, embeddings) to questions that do not reduce cleanly to a table
- Design the evaluation for your own work: baselines, gold sets, agreement measures, honest error analysis
- Work with our medical officer to keep the modelling grounded in how medicine is actually practised
What we are looking for
- 3 to 5 years of applied data science, ML or quantitative work
- Genuine strength in mathematics, statistics and formal logic. You should be comfortable reasoning about graphs from first principles, not just calling a library
- Working knowledge of graph theory and its practical use: representation, traversal, connectivity, similarity, and what breaks as graphs grow
- Strong Python and SQL, and the general engineering competence to get data in and out of things reliably
- Rigour about measurement. You establish a baseline before you claim an improvement, and you can defend your numbers
- Clear written reasoning. You can explain a method to someone who is not a data scientist
Helpful, not required
- Experience with clinical, biomedical or other ontology-heavy data
- NLP experience: entity recognition, relation extraction, embeddings, retrieval
Why this role is interesting
South African clinical data is messy in specific ways: multilingual consultations, code-switching mid-sentence, a fragmented mix of public and private care, and terminology standards that were not designed here. The problems are real, they are not solved, and the work goes into production in front of doctors rather than into a slide deck.
To apply
Email a short note and your CV or LinkedIn profile to hello@akiliai.net with “Data Scientist (Graph)” in the subject line. Tell us about a problem you modelled as a graph and what you got wrong the first time.
Apply by email