Geneva AI and Scientific-Discovery PhDs: Research-Fit Application Guide
Prepare for Geneva PhD searches in AI foundation models and scientific discovery: research fit, reproducible projects, domain knowledge and vacancy-status checks.
By Prime Scholarship Alerts Editorial
A search for Geneva PhD positions in foundation models and scientific discovery combines two requirements: machine-learning preparation and the scientific problem the project aims to solve. The University of Geneva's AI Science Center presents interdisciplinary research across scientific domains, including FUNDIS. A centre or project page is useful background, but it is not by itself a live recruitment notice.
Check the individual position's status
The cited FUNDIS astrophysics listing on the European Astronomical Society directory gives a closing date of 31 August 2026. That date has passed as of this guide's 7 September 2026 review. Do not treat that specific call as open without a newer official notice. Other disciplines or future rounds can have different dates and eligibility.
Record the project, department, vacancy reference, closing date and official application destination for each position you find. Avoid assuming that a broad announcement of several positions means every research group accepts the same documents or remains open for the same period.
Start with the scientific question
Write a plain-language description of what the research is trying to learn. Then identify how modelling, data or simulation contributes to that question. For an astronomy project, general enthusiasm for chatbots does not demonstrate the same preparation as understanding the relevant observations or inference problem. Be explicit about the domain knowledge you already have and what you need to develop.
Read one or two closely related papers carefully. Note the data, comparison methods, limitations and unanswered questions. Your goal is not to invent a complete thesis before applying, but to show that you can engage with the project's actual research rather than repeat its title.
Present one reproducible project
Choose work you can explain from data preparation to evaluation. Describe your own contribution, the comparison you used and what the result does not establish. If the work was a class assignment or team project, say so. Explain how someone could understand or reproduce the permitted parts without accessing confidential data.
A portfolio note can include the question, dataset origin, experiment design, evaluation and limitations. Keep the technical detail proportionate to the call. Do not claim that a high score proves scientific usefulness without considering the task and comparison. Use genuine code and results, and acknowledge collaborators.
Prepare for a research conversation
Practise explaining a failed experiment and your next diagnostic step. Be ready to discuss why a baseline matters, how you kept evaluation separate from development and where your experience is limited. A thoughtful account of uncertainty is more useful than claiming expertise in every model family listed in an advert.
Match the funding and application documents
Check the required degree, transcripts, statement, references and funding terms in the specific call. Confirm whether the appointment and doctoral admission are separate processes. Keep your research interests consistent across the CV and motivation statement. Save submission confirmation, and never describe an expired call as current simply because its webpage remains online.
Official sources
Related preparation guides
Frequently asked questions
Does the AI Science Center page confirm an open PhD?
No. Use it to understand research areas, then verify a specific current vacancy and its application instructions.
Can I apply with machine-learning experience but limited domain knowledge?
Check the particular call. Explain your relevant preparation honestly and identify the scientific knowledge you would need to develop; do not assume every project accepts the same background.