Private courses for your team
We run courses privately for your team, using your datasets and the problems your people work on. Examples built on your own data are the ones participants keep using after the course.
Each trainer takes up to twelve participants. In a group that size, the trainer can look at the code on your screen and help you fix it, which a recorded video cannot do. For larger groups we bring additional trainers, so the ratio holds however many people you send.
Any of our courses can run this way. We also design new ones. We start from what your team already knows and what they need to be able to do, and build the course to cover the difference. If a subject is outside our expertise, we bring in specialists and work with them.
Some courses need groundwork first. In that case we take on consulting work to understand your systems and develop the material.
Get a quoteWhy teams book privately
- Up to 12 participants per trainer. Send a bigger group and we add trainers to keep the ratio.
- Your data. On request we teach with your datasets and code bases, so your team recognises the worked examples.
- One shared approach. The whole team learns the same methods, and senior staff help juniors while learning themselves.
- One booking. One date and one invoice, instead of booking seats one at a time across a dozen calendars.
- Your schedule. At your office or online, and split across days or weeks if that suits the team.
Better than a team-building day
A team that spends two days solving real problems together gets what an offsite is meant to deliver, and leaves with working code as well. Colleagues who rarely compare notes often find they have been solving the same problem twice, in different ways.
Mixed experience levels help. Senior staff test their own understanding by explaining it, and junior staff ask questions a set syllabus would not anticipate. Everyone leaves with the same vocabulary, and the same worked examples to refer to in code review.
Private training also works across teams. Analysts and engineers who take a course together tend to stop handing each other work in formats the other has to undo.
How it works
Tell us what you need
A short conversation about what your team does now, what you want them doing, and how much Python they already write.
We design the course
You get an outline to review before anything is booked, built from our material and yours. Change it as much as you like.
We run it
At your office or online, on dates that suit you. Participants write code in every session, rather than sitting through slides.
Afterwards
Participants keep the course materials and notebooks, and can bring questions to our open office hours.
Popular custom training topics
Machine learning foundations
Supervised and unsupervised learning, feature engineering, and model evaluation. The emphasis is on knowing which method fits a problem and why. Your team can then rule options out quickly, instead of trying everything scikit-learn offers and keeping whatever scored highest on one split.
Deep learning
Neural networks with PyTorch: computer vision, sequence models and transformers. We build up from the training loop rather than assembling copied snippets. This topic includes when a deep model is the wrong answer, and a simpler one will hold up better in production.
Building with LLMs
Retrieval-augmented generation, fine-tuning, and the engineering around a language model in a real application. This covers evaluating output that differs on every run, and the failure modes that only appear once real users start asking questions.
AI-assisted coding
Getting real work out of coding assistants, and reviewing what they produce. The skill that matters is judging generated code you did not write, quickly and honestly. Your team ships faster without building up code nobody understands.
Putting models into production
Deployment, monitoring and drift detection: the engineering that keeps a model working after the notebook is closed. This covers what to log, how to tell degradation from noise, and how to retrain without breaking what already depends on the model.
Evaluation and model risk
Measuring whether a model is any good, and where it fails. This topic is for teams who have to satisfy a reviewer, a regulator or an internal risk function. They need evidence that holds up when someone else reads it carefully.
Something not listed here
This field changes faster than any course list. The question your team has may not have had a name last year. We teach it at the depth you need, and where the expertise sits outside our own, we find the people who have it. Tell us the problem.
Every course here started as someone's specific problem.
Discuss your requirementsTalk to us about your team
Tell us what your team works on and how many of them there are. We will come back with an outline and a fixed price, usually within two business days.
Request a quote