The Aperintel AI Academy is the training arm of Aperintel, an AI engineering company shipping production AI systems for paying clients across fintech, education, regulated industry, governance, and community platforms.
Most software education teaches people to generate code. The Aperintel AI Academy teaches them to architect, ship, and own production AI systems in partnership with the machines that now write the syntax. That is the load-bearing claim.
Aperintel's shipped products prove the curriculum is real. The Academy's graduates become Aperintel's contractor pool. The two surfaces compound on each other rather than competing for attention. Aperintel charges day rates and project fees on the engineering side and cohort fees on the Academy side. The Apprenticeship is the strategic bridge between the two.
The curriculum is grounded in products Aperintel has shipped or is shipping: Marktrader (multi-factor financial AI), TekkieStack (offline-first AI literacy for children), the Aperintel AI Gateway (multi-model routing with cryptographic audit), Nexuscone (the open-source audit primitive at github.com/nexuscone/nexuscone), Rallova (multi-tenant community SaaS), Aces (diaspora community platform), Collstack (audit-grade academic operations), and Metacarpal (the personal autonomous-agent operating system that runs underneath much of Aperintel's internal work).
Curriculum content nobody else teaches properly, taught across multiple tracks at different depths.
The structured approach to writing prompts that produce reliable, audit-friendly, anti-cliché output. Context first, instruction second, constraints third. Structured output as the default.
The long-form markdown context file every learner builds in Foundation Week 2 and refines across the rest of the curriculum. Working role, projects, technical stack, working preferences, constraints, what the learner wants from the LLM.
The first commit of a new project: README as ADR, LICENSE, .gitignore, .env.example, folder structure, CLAUDE.md / AGENTS.md, account creation discipline, audit-chain consideration from day one.
Audit-friendly systems by design using the open-source Nexuscone primitive. Hash chains, canonical JSON, optional Ed25519 signing, OpenTimestamps anchoring when v0.2 lands.
Daily practices that keep a long-running codebase healthy. Branch hygiene, dependency hygiene, documentation hygiene, secret hygiene, the Friday housekeeping hour.
The catalogue of how AI-assisted systems fail in production, with a runbook for each. Hallucinated API endpoints, runaway agentic loops, cost spikes, prompt injection, model deprecation events, training-data updates that change behaviour silently.
Ninety minutes, live. One real Aperintel problem walked through end to end. A practical demo of the Prompting Protocol. A working template you keep. £100 discount code on the next Foundation cohort.