3 h
Weekly group session
One working session per week, five participants maximum: each project's progress is presented, the trainer reviews code live, and everyone learns from the choices — and mistakes — of the four other projects.
Learn to build AI solutions with Claude Code — and leave with your own project, carried end to end.
4 intensive weeks · cohorts of 5 participants maximum
Very small cohorts and intensive individual coaching: with five participants, nobody stays a spectator, and every hour of supervision is spent on your real project.
3 h
One working session per week, five participants maximum: each project's progress is presented, the trainer reviews code live, and everyone learns from the choices — and mistakes — of the four other projects.
2 × 1 h
Two one-to-one meetings with the trainer every week: concepts tailored to your project, guided review of your code, hard points unblocked in session, and a sustained pace of progress.
4 sem.
No artificial labs: each week's playing field is your own project, scoped in week 1 and delivered at the end of week 4, aligned from the start with the CASD™ certifying-project specification brief.
Each week combines the group session, two individual meetings and progress on your project, following the six domains of the official CASD™ curriculum.
Claude models, costs and selection; Messages API, streaming and error handling; capstone scoping workshop against the specification brief.
Deliverables: 1-page scoping note · initialised repository · working API integration
Structured prompts and test sets, robustness to injection; from single call to tool-equipped agentic loop, bounds and budgets.
Deliverables: versioned prompts · tested tool-equipped agent
An MCP server built and secured; a RAG chain with sourced answers; measured prompt caching and a cost strategy.
Deliverables: MCP server · RAG component · caching report
Hardening, evals and observability, responsible use; mock MCQ under real conditions and an individual mock defence.
Deliverables: complete project (code, report, video) · mock MCQ · mock defence
Personal work: about 5 to 8 hours per week, spent on your project between sessions. Prerequisites: 2 years of professional development experience, Python 3.11+ or TypeScript/Node.js 20+, your own Anthropic API key and a project idea — checked in a 30-minute placement interview.
The program prepares you for the three components of the CASD™ assessment — and your capstone, aligned from week 1 with the official specification brief, can be submitted as-is as the certifying project.
60 %
Built against the public specification brief, assessed independently by two SGCB examiners — never by your trainer.
25 %
Live demonstration, technical questions and authorship verification; an individual mock defence is held in week 4.
15 %
A knowledge questionnaire covering the six domains of the framework; a mock MCQ under real conditions is taken and debriefed in week 4.
Independence of the certification. The CASD™ certification is awarded by the Sopheria Global Certification Board, independent from the Academy. Registration for the assessment is separate, no grade from the training is passed to the board, and completing the program confers neither exemption nor advantage: the assessment is identical for all candidates.
Next cohort: September 2026 — enrollment closes as soon as the cohort of five is full.
Sopheria Global Academy · Training
CASD™ certification not included: the assessment is taken with the Sopheria Global Certification Board, independently of the training.
Program fee Pricing on request
Secure payment by card or mobile money · Invoice available · The SGA trains, the SGCB certifies.