Skip to main content

Technical Practice and AI Literacy

The orientation now includes three practical technical modules, each with three concept clusters, six lessons, three practice tasks, worked examples, a runnable project, case studies, mistake clinic, selective resources and an independent assessment.

New skills and artifacts​

  • Module 5: Data, Files and Automation, 14–20 hours: bytes/text, schemas, CSV parsing, staged validation, grouped reports, fingerprints and reproducible commands. Build a CSV-to-JSON report tool.
  • Module 6: The Web, HTTP and APIs, 16–24 hours: URLs, messages, response contracts, failure classification and actual local exchanges. Build an inspectable API client using a supplied teaching server.
  • Module 7: AI Literacy and Assisted Engineering, 12–18 hours: model/context/tool distinctions, focused assistance, tutor/debug/review workflows, independent evaluations and untrusted-content boundaries. Build an evaluator for plausible suggestions and demonstrate unaided transfer.

These add 42–62 hours. With learning support and existing technical work, Semester 00 is now 210–309 focused hours. Estimates await learner pilots; they are not promises about completion speed. Workload and Progression gives the non-overlapping totals.

Prerequisites and order​

Complete the immediate Programming Basics gate before the implementation labs. While its delayed checks are pending, you can do independent orientation work. Module 5 uses its functions, collections and file boundaries. Module 6 follows the representation/schema lessons and uses a provided transport adapter. Module 7 follows basic Python and uses Semester 00 Module 4 learning calibration and the Launchpad reading/evidence and no-code AI modules.

A workable interleaving is Module 0 → Module 5 with Algorithm Intuition → Module 6 with CS Fundamentals → Module 7 with Code Readability and review. Do not postpone all practical work until after every nontechnical module; use the support skills inside these projects. The module numbers identify sidebar groups, not an instruction to finish every page numerically before writing code.

One downloadable toolkit, three distinct contracts​

Download Orientation Toolkit. Extract once. It contains starter/reference implementations, checks, CSV and policy fixtures, a CLI, a local transport adapter and a disposable server. Python 3.9+ standard library is sufficient. No paid account, API key, cloud deployment or dependency installation is required.

Module 5 implements import/report behavior. Module 6 implements response interpretation and verifies a real loopback exchange. Module 7 implements bounded declarative policy evaluation; its sample suggestions are authored fixtures, not actual model outputs. Optional live AI work must be labeled and independently verified. Offline completion demonstrates model literacy and evaluation practice, not provider-specific operational expertise.

A server-binding restriction leaves the Module 6 live gate pending; fixture checks do not substitute for an actual exchange. The lab starts no public service and validates no TLS, public DNS, authentication or general Internet behavior. No generated command is executed by the AI evaluator.

Assessment and evidence reuse​

Require DF1–DF4, WH1–WH4 and AI1–AI4 at 2/2 under the respective module rubrics, in addition to all existing gates. The exam retains its four equal parts and existing score floors; technical practice cannot compensate for a failed original outcome. A student with equivalent skills may demonstrate fresh transfer and evidence instead of repeating basic instruction.

Keep one toolkit folder with per-module implementations, predictions, test transcripts, failure repairs and independent transfer. Reuse existing Git history, assistance declarations and privacy inventory. New technical hours cover new lessons, implementation and assessment. Existing AI/privacy strand hours still cover declaration and ethical/privacy/accessibility reasoning; do not charge the same log-writing activity twice. Existing programming retention checks keep their own time and requirements.

At the semester checkpoint, inspect the three already-assessed lab packets; this brief inspection stays inside existing checkpoint time. The Algorithm Explorer project remains the integration project. Do not add a second full application merely to combine all three labs. A short explanation of how bytes become records, responses and evaluated suggestions is sufficient to connect them.

Why these topics, and what comes later​

Terminal and Git are already taught in Launchpad. The CS overview already names computing areas. These additions instead require observable technical work: importing imperfect data, communicating with another process and evaluating assistance against a contract.

Hardware organization, OS internals, full networking, database transactions, production API compatibility, ML training and autonomous agent deployment remain in their later semesters. Introductory exposure here prepares vocabulary and testable habits without claiming those subjects are complete.

Research and selective resources​

Sources checked 2026-10-27. Full-book reading is not required. Our lesson sequence, workloads, fixtures and assessment thresholds are original authoring decisions awaiting independent review and learner calibration. Passing executable checks verifies selected software behavior, not educational effectiveness or live AI reliability.