Artificial intelligence for scientific research
Search, read and write with AI, then train and evaluate your own model — without writing code.
| Duration | 2 jours (14 h) |
|---|---|
| Format | On-site |
| Level | Beginner |
| Participants | — |
| Price | Sur devis |
| Certificate | Yes |
Who is this course for ?
Doctoral students, researchers, teaching researchers, engineers and laboratory professionals. No programming prerequisite.
What will you be able to do by the end ?
- Put an artificial intelligence to work on your own documents, with sourced answers.
- Keep unpublished data out of any online service, knowing which solution to install.
- Turn reports, tables and laboratory notebooks into usable data, in bulk.
- Train a model on your own measurements or images, without writing code.
- Judge whether a result is valid, and describe it correctly in a thesis or a paper.
Programme
-
Day 1 — Beyond the chat window
The three levels of use: chatting, grounding on your own sources, training your own model. Why a language model invents, and what that means for a bibliography. Grounding AI on your corpus: sourced answers, structured synthesis, concept map.
-
Day 1 — Your documents, your sensitive data
Demonstration of an AI installed on a machine, answering from confidential documents without any data leaving it. Bulk extraction: analysis reports and handwritten notebooks turned into verified tables. Automating a repetitive task without code. Research integrity, law 18-07 and decree 25-320.
-
Day 2 — Training your first model
Recognising problems that call for machine learning. Training a classifier on your own images — microscopy, samples, plates — without writing a line of code, observing its errors and correcting your data.
-
Day 2 — Measuring, validating, publishing
Principal component analysis, classification and regression on laboratory data, by drag and drop. Train-test separation, data leakage, overfitting, honest metrics. Sharing a demonstration of your model. Individual action plan.
Prerequisites
No programming prerequisite. One laptop per participant. Participants are invited to bring their own data: measurement tables, spectra, images, documents.
How is the course run ?
Two days in which every sequence produces something the participant takes away: a searchable corpus, a table extracted from documents, a trained model. Workshops are based on the data participants bring, collected through a questionnaire before the session.
Upcoming dates
-
Open for registration · Organised by CRAPC Learning
Further reading
-
What machine do you need to run an AI locally?
Video memory decides almost everything. Here is how to size a machine to run a language model on your premises, and how to check for yourself before buying.
-
How do you make an AI answer on your own documents?
A language model does not know your procedures. The method that works is to give it the right extracts at the moment of the question, rather than trying to teach them to it.
Frequently asked questions
Are courses delivered at our offices?
Yes, and that is the format we recommend: your teams work in their own environment on their own data. Remote and hybrid delivery are possible depending on your constraints.
How many participants per session?
We deliberately cap group sizes to keep time for individual support during exercises. The exact number is set in the quote, according to the course and the level of the group.
Do we work on our own real data?
Yes wherever possible, and this is the main difference from generic training. Exercises use your real datasets, which means participants leave with dashboards and automations they can put straight to use.
Is a certificate issued?
Yes, a named certificate is given to each participant at the end of the course, together with the detailed programme covered.
What happens after the course?
The materials, exercise workbooks and working files stay with you. Follow-up support can be included in the quote, to answer the questions that surface once the learning is put into practice.
How much does a training course cost?
The price depends on four things: the course chosen, its length, the number of participants and the location. An in-house session run on your own data is not priced like a short refresher. We send a quote within 48 hours, free and with no commitment, after a thirty-minute conversation to frame the need and the level of the group.
Updated Oct. 7, 2026