Pi Sandbox: A 16-Session Learning Plan

Following on from my initial sandbox setup, I’ve now developed a comprehensive project plan with the help of Claude Code. The goal: refresh my physical computing and engineering mathematics skills while building a properly documented portfolio of projects.

The Problem I’m Solving

I’ve done a lot of this work before - Python maths, GPIO programming, sensors - but documented it poorly. This time I want to:

• Create well-structured, reusable code
• Build a knowledge base with linked notes (using ObsLite)
• Document the learning process for future reference
• Work at a sustainable pace (~3 hours/week)

What I’m Working With

Hardware

• Raspberry Pi 5 - Main development machine
• Sunfounder Super Kit V2.0 - LEDs, sensors, displays, motors
• micro:bits - Wireless sensor nodes
• Arduinos - Analog sensing (Pi has no ADC)
• Mac Mini M4 Pro - Heavy compute for ML work
• Ender 3 + OctoPi - 3D printing for enclosures

Existing Resources

During the planning session, Claude explored my existing codebase and found:

• 60+ Python maths notebooks covering SymPy, NumPy, SciPy
• ObsLite - A markdown editor I built with Tauri/Svelte/Rust
• 14 Sunfounder lessons with Python code and Fritzing schematics

The 16-Session Plan

Phase 1: Foundation (Sessions 1-2)

• Environment setup: VSCode, venv, Jupyter
• Hardware Catalog: A living document of all components for project planning

Phase 2: Physical Computing (Sessions 3-5)

• Pi GPIO essentials (curated, not all 14 tutorials)
• micro:bit as wireless sensor node
• I2C sensors (accelerometer, LCD display)

Phase 3: Engineering Mathematics (Sessions 6-10)

This is the core of the project - building 5 polished notebooks covering 1st year engineering maths:

Notebook	Topics	Tools
01_calculus	Differentiation, integration	SymPy, NumPy
02_differential_equations	RC circuits, oscillations	SciPy, Mathematica
03_linear_algebra	Circuits, statics, eigenvalues	NumPy linalg
04_statistics	Sensor data analysis	SciPy stats, Pandas
05_mathematica_comparison	When to use which tool	Wolfram Language

Each notebook will include:

• Theory with worked examples
• Engineering application
• Python + Mathematica comparison
• Exercises with solutions

Phase 4: Applied Projects (Sessions 11-13)

• Sensor + maths integration (FFT, filtering)
• Control systems introduction
• 3D printed sensor mounts

Phase 5: ML & AI (Sessions 14+)

• PyTorch on Mac Mini
• Local LLMs with Ollama
• Capstone projects

The Workflow

Session Structure (~3 hours)

0:00 - 0:10  Review SESSION_LOG, check plan
0:10 - 0:30  Setup hardware, activate venv
0:30 - 2:30  Main work
2:30 - 2:50  Document, commit with clear messages
2:50 - 3:00  Write "Next Steps" for next session

Documentation Stack

• Code: VSCode on Pi with Jupyter extension
• Notes: ObsLite with wiki-links
• Blog: Posts here to Design_Art_Research
• Version Control: Git with explanatory commits

Key Planning Decisions

1. VSCode runs on the Pi - not Remote-SSH from another machine
2. Curated sensors - skip 7-segment, dot-matrix, NE555
3. micro:bit priority - great for wireless sensor networks
4. Mathematica included - it’s free on Pi, why not use it?
5. Hardware catalog - a living reference for project planning

Next Steps

Session 1 will focus on:

1. Installing/verifying VSCode on Pi 5
2. Creating the project venv
3. Configuring Jupyter
4. Setting up ObsLite vault
5. Running the hello_world notebook

The full project plan and session logs are in the pi-sandbox repo.

---

This planning session was conducted with Claude Code, which explored my existing codebase, identified resources, and helped design a realistic learning plan.