SCHOOLREGISTRY ACADEMY · COURSE GUIDE

Applied AI Builder Certificate

A ten-week certificate course in applied AI, taught inside Larkhollow Studio, a fictional twelve-person AI studio in Austin with a Lagos desk. You take three client accounts for eight weeks: a vet group that books appointments and triages, an outfitter answering product questions across a nine hundred item catalogue, and an accounting firm taking in documents. You write prompts as products with versions and change notes, ground them in a knowledge base so the assistant stops inventing prices, build evaluation sets from ground truth rather than from your own prompt, connect tools with an approval switch on anything touching money, health, law or admission, and watch the cost. The line the whole course is built on: the model drafts, a person decides.

Start: October 12, 2026 · 11 AM WAT (Nigeria).
Fee: ₦200,000 for this track, or two ₦100,000 installments with the balance due three weeks later.
Format: Ten weeks: eight of lessons, assignments and one live class a week, then two practicum weeks of timed shifts on the track's job simulator. Your track’s live-session day and time are confirmed at enrolment.

What you will learn

Your curriculum

Week 1: What an applied AI builder does, and for whom

The studio and its nine client accounts; what a language model is and is not; the builder's line, the model drafts and a person decides; the three accounts on your seat; the bench and the prompt library; the Lagos and US lanes.

Week 2: Prompts as products

Instructions, examples, output format and refusal rules; why wording drifts and how you pin it; examples that teach against the one that misleads; versions, diffs and change notes; writing for Nigerian customers, naira, names and the polite decline.

Week 3: Grounding

What a knowledge base is and what belongs in it; documents, chunks and retrieval in plain words; answering from the base or saying you do not know; the catalogue that will not fit in a prompt; the invented price and the three-part fix.

Week 4: Evaluation

The ground-truth set and its ten question types; sampling by type so the eval looks like the traffic; the eval set written from your own prompt and why it is worthless; regression runs before a change ships; reading a failure report.

Week 5: Tools and actions

What a tool call is; connecting a calendar, a sheet and a ticket queue; permissions you can read back as read, draft or write; the approval switch on money, health, legal and admission; bookings that stay draft; the request to remove the switch, refused.

Week 6: Running it

Calls, tokens, cost and latency; the bench rate and what a month costs against the client's budget; alerts before the invoice; the weekly usage review with the client; the change log an account can be handed over on.

Week 7: When it goes wrong, and it is yours

The wrong price reached a hundred and forty customers; who you tell and in what order; bad news reported as it is; the post-mortem that changes the eval set; personal data that should never have been loaded; the automation you refuse to build.

Week 8: Working the job

Three assistants with their eval reports, one post-mortem and one refusal memo; the handover file; the timed practice exam with a readout by outline area; the Lagos and US markets honestly; interview answers from your own work; the practicum brief.

Prepare to participate

Plan for a computer, headset, reliable internet, a quiet practice space, and time for assignments. Review the student handbook for assessment, access, and policy details before payment.

Certification

Issued by SchoolRegistry Academy when the track’s assessment requirements are met. It is not third-party accreditation and does not promise job placement.

View the course and enroll ↗

Questions: hello@schoolregistry.ng

Use Print → Save as PDF for a PDF copy. Course guide · September 2026.