After this module, read the Cali accident (AA965, 1995) — an FMS database error and crew confusion. Ask: what did the system make easy vs hard? That's a safety analysis question, not just a technical one.
This module directly feeds AF447 analysis. The crew didn't understand what mode the automation was in. After learning autopilot modes, re-read the AF447 BEA report. The technical understanding will completely change how you read it.
CFIT (Controlled Flight Into Terrain) was the #1 fatal accident cause before EGPWS. Look up the Avianca 052 (1990) or Korean Air 801 (1997) accidents to see what GPWS could and couldn't do. The safety lesson: a warning system only works if the crew responds to it.
Entry point. Start here. Ignore everything else until this is done.
Weeks 1–4 · ~45 min weekdays, 2 hrs weekendsDraw a Safety System Map for a hypothetical airline. Show 3 layers: organizational (management, policy, culture), technical (aircraft, maintenance, IT), human (crew, ATC, dispatchers). Overlay SMS pillars. Mark where reactive/proactive/predictive safety lives. Add one example failure in each layer. Hand-drawn is fine. This is your first artifact.
Pick one accident. Stick with it. Use it for everything in this block.
Weeks 5–8 · Recommended: Air France 447 (BEA report) or Colgan Air 3407 (NTSB report)For your chosen accident: (1) Causal chain diagram — events → conditions → latent failures. (2) Bow-Tie diagram. (3) Rewrite the official "probable cause" without blaming individuals. (4) Write 3 system-level safety recommendations. All in a single document. Push to GitHub or keep in a portfolio folder.
This is where CE engineers separate themselves from everyone else in safety.
Weeks 9–12 · Your CS background is your advantage hereUsing NTSB accident data (free CSV at ntsb.gov): plot fatal accident rate per million departures over time. Distinguish signal from noise using a control chart. Show how the same data can support opposite conclusions depending on how you cut it. Write a 1-page critique: "Why this chart might be lying." This is your first real data analysis piece.
You already know Python. This is applying it to a domain. Should feel fast.
Weeks 13–16 · pandas + matplotlib + real NTSB dataGitHub repo: aviation_safety_db. Contains: ETL script (raw CSV → cleaned → SQLite), schema diagram, query file with 10 safety questions answered. README explains what safety question the project answers. This is your first public portfolio piece. Make the README good.
Re-analyze your Block 2 accident using HFACS (full table) and TEM (threats/errors/UAS breakdown). Write a 1-page human factors narrative about the crew — without using "mistake", "negligence", "forgot", or "failed to". Every action must sound understandable given the constraints they faced. This is the hardest thing you'll write. Do it anyway.
Everything feeds into this. This is what you show at interviews and on applications.
Weeks 21–28 · Live deployed app + professional safety reportA safety manager opens your dashboard and asks: "What's our biggest risk in the approach phase, and what data supports that?" Can you give a defensible, data-backed answer in under 3 minutes? That's the bar. Everything in this plan builds toward that moment.