Hands-on Projects

Five projects.
Increasing complexity.

Every module ends with something you build and keep. The projects form a deliberate arc: from a single HTML file in a browser to a deployed multi-user application — mirroring the progression through the Five Levels.

P01 RC Filter Analyser Level 3–4 P02 CLI Tool with File Integration Level 4 P03 Serial Data Plotter GUI Level 4 P04 Datasheet Q&A Tool Level 4–5 P05 Resource Booking System Level 5
P01
Module 1 — Project 1 Level 3–4

RC Filter Analyser

Interactive browser tool with live Bode plot

HTMLJavaScriptCanvas APINo install

Build an interactive RC filter analyser entirely inside the Claude browser interface. No editor. No terminal. No install. You write a specification, hand it to Claude, and receive a complete, runnable HTML file.

What you'll learn

  • Write a precise Level 4 specification covering task, context, constraints, and output format
  • Use the Canvas API for real-time Bode plot rendering (magnitude and phase)
  • Apply the four-element framing technique to a physics-based problem
  • Evaluate and iterate on AI output without leaving the browser
RC Filter Analyser running in the browser — Bode plot updates live as R and C sliders change Module 1, A1.1 · Project 1 demo · Built entirely in Claude browser interface

Technical detail

The tool calculates the cutoff frequency fc = 1/(2πRC), renders a magnitude response in dB on a logarithmic frequency axis from 1 Hz to 1 MHz, and updates the plot in real time as the user adjusts R and C sliders. All in a single self-contained HTML file — no external dependencies.

P02
Module 3 — Project 2 Level 4

CLI Tool with File Integration

Process maker file formats from the command line

PythonCLIargparseCSV / JSON

Build a command-line tool that reads and processes structured files from your hardware workflows — KiCad BOMs, component exports, measurement CSVs, or serial trace data. Cline or Claude Code does the heavy lifting from a detailed specification.

What you'll learn

  • Structure a multi-file Python project with an agent coding assistant
  • Write specifications for file-processing logic: parsers, validators, formatters
  • Use argparse for a clean CLI interface
  • Handle real-world file format quirks (encoding, delimiter variation, missing fields)
CLI tool output — processing a KiCad BOM CSV and producing a formatted cost and availability report Module 3, A3.3 · Project 2 demo · Python CLI with argparse

Technical detail

Project candidates include a BOM cost estimator (reads KiCad CSV, queries vendor pricing), a logic analyser trace decoder (reads .logicdata or CSV, formats for human inspection), and a power budget calculator (reads component specs, sums draw, flags risks). The exact project is confirmed at course launch.

P03
Module 3 — Project 3 Level 4

Serial Data Plotter GUI

Live web dashboard for Arduino sensor data

PythonFlaskChart.jspyserialWebSocket

Reads multiple sensor channels from an Arduino over serial, streams the data to a live web chart, and logs everything to CSV. A real multi-file project built with Cline or Claude Code acting as the primary implementation agent.

What you'll learn

  • Direct an AI agent across a multi-file Python + web project
  • Use Flask-SocketIO or Server-Sent Events to push data from serial port to browser
  • Implement live Chart.js plots with configurable channel colours
  • Design the specification workflow for a project with hardware dependencies
Serial Plotter GUI — three live channels from an Arduino plotted in the browser with CSV logging Module 3, A3.5 · Project 3 demo · Python + Flask + Chart.js

Technical detail

Serial data arrives as comma-separated lines: timestamp, and one or more sensor channels. Flask serves the web UI and streams updates via WebSocket. Chart.js renders all channels on a shared time axis. On close, accumulated data writes to a timestamped CSV. The Arduino sketch is provided.

P04
Module 4 — Project 4 Level 4–5

Datasheet Q&A Tool

Natural-language questions over any PDF datasheet

PythonRAGPDF parsingLLM APIVector search

Upload a component datasheet, ask natural-language questions, get answers with citations back to the source page. Your first application with an LLM at runtime — introducing retrieval-augmented generation (RAG) at the simplest useful level.

What you'll learn

  • Call the Claude API from Python (or an equivalent) with cost and latency awareness
  • Parse and chunk a PDF for efficient LLM ingestion
  • Build a simple vector search index for retrieval
  • Design a RAG pipeline: chunk → embed → retrieve → generate → cite
Datasheet Q&A in action — asking a natural-language question, getting a cited answer from a PDF datasheet Module 4, A4.1 · Project 4 demo · Python + RAG + LLM API

Technical detail

The application ingests any PDF (component datasheets, application notes, standards excerpts), chunks it by page or paragraph, embeds the chunks, and retrieves the top-k most relevant chunks for each query. The LLM generates an answer grounded in the retrieved text and cites the source pages. Runs locally — no cloud vector database required.

P05
Module 5 — Capstone Project Level 5

Resource Booking System

Full-stack makerspace equipment booking application

PythonFlaskPostgreSQLSQLAlchemyAPSchedulerAuth

The capstone: a production-grade multi-user web application for booking makerspace equipment. Built with Claude Code running as an autonomous agent across ten-plus files, with you steering the architecture and reviewing every diff.

What you'll learn

  • Direct a Level 5 agent across a complex, multi-file project
  • Design a PostgreSQL schema with ORM mapping (SQLAlchemy)
  • Implement role-based access control (admin, member, guest)
  • Integrate an external API (Nager.Date for public holidays)
  • Build background jobs with APScheduler for reminder emails
  • Practice the review-and-steer discipline that makes Level 5 safe to use
Resource Booking System — the capstone running locally, showing the booking calendar and conflict detection Module 5, A5.6 · Capstone demo · Python + Flask + PostgreSQL

Technical detail

The application manages equipment (laser cutters, CNC machines, 3D printers), booking time slots, users and roles, booking conflict detection, public holiday blocking via Nager.Date, and APScheduler email reminders 24 hours before each booking. Built with Flask, PostgreSQL, SQLAlchemy, and Flask-Login. Deployed locally in development; deployment guides for VPS hosting are included.

Get Started

Build all five. Keep all five.

Each project is yours to extend, modify, and ship. They're designed to be real tools — not tutorial throwaways.

Enroll Now Join the Email List