Training · Founding cohort

Use AI in your own work, and know when it is wrong

Eight weeks, two hours a day, five days a week — live, with an instructor and a small group. For professionals who want to put these models to work inside their own practice: privately, on their own machine, over their own documents. The part nobody else teaches is the part that matters most — how to tell whether the answer you just got is actually correct.

Who this is for. Doctors, dentists, lawyers, architects, engineers, pharmacists, accountants, researchers, writers, film-makers, builders — anyone whose work involves documents, judgement and confidentiality, and who has watched two years of AI claims without a way to separate the useful from the noise.

Who this is not for. If you want to train and serve models yourself — LoRA, quantization, eval harnesses — this is the wrong course. That is a separate cohort we have not scheduled yet; write to us and we will tell you when it opens. And if you are looking for a certificate to put on a CV without doing the work, we would rather you did not enrol: the capstone is the qualification and it cannot be faked.

The eight weeks

WeekWhat you learn
1What these models actually are, and what they cannot do. Reading a model board: price, context, licensing, and who owns what you put in.
2Running a model on your own machine. Why local matters when the documents are client-confidential, and what your hardware can and cannot handle.
3Your own documents. Retrieval, chunking and citations — making a model answer from your files instead of from its memory.
4Judging the output. Building a rubric for your own field and testing against it. Where hallucination comes from and how you would catch it before a client does.
5Workflow. Intake, drafting, summarising, review — automating the parts of your practice that are repetitive without automating the parts that need you.
6Privacy, confidentiality, and the regulatory line in your profession. What you may not put into a hosted model, and how to work anyway.
7Going further: prompt libraries, structured output, and an honest answer to when a custom model is worth paying for — usually later than you would think.
8Capstone. A working workflow for your own practice, and a written account of where it fails.

What you leave with

Format

Length8 weeks · 2 hours a day · 5 days a week · 80 contact hours
LiveInstructor-led, with 1–2 assistants supporting the lab work
Cohort sizeCapped at 20, so questions get answered
You needA laptop with 16GB of memory or more, and the willingness to type commands. No prior programming.
RecordingsEvery session recorded and yours to keep

Fees

Founding cohortStandard
India₹35,000₹55,000
International$650$900

The founding cohort is priced below what the course is worth, deliberately and only once. In exchange we ask for an honest testimonial at the end and permission to publish your capstone as a case study. We are early enough that your results are worth more to us than your money, and it seems better to say that than to invent a discount reason.

What the certificate means

It attests that you attended and that your capstone met the published rubric. That is all it attests to. It is not an accreditation, it is not a licence to practise anything, and it is not a job guarantee — nobody is in a position to guarantee you a job and you should distrust anyone who says otherwise. What actually carries weight is the capstone itself, which is why it is the part we grade.

What we will not claim

Applying

Email contact@naderu.com with your profession, what you would want your capstone to be, and what you have already tried with AI in your work. We read those properly — the cohort is capped, and a group that wants roughly the same thing learns faster than a group that does not.

Not sure yet? The model board and the weekly reports are free and public. They are the same habits of mind this course teaches — every number sourced, every claim dated, corrections published when we get it wrong. Read a few and see whether you want eight weeks of it.