AI in Aviation

A collection of Jupyter notebooks demonstrating practical applications of artificial intelligence (AI) in the airline and aviation industry. Each notebook is accompanied by a Medium article that explains the airline/aviation news and AI concepts.

Contents

• 10 Mar 2026 Airspace Closure Impact Forecasting with a Lightweight LSTM ✈️ We train a small 2-layer LSTM in PyTorch to forecast the next 6 hours of UAE/Gulf hub flight cancellation rates from the prior 24 hours using synthetic airspace-closure recovery episodes. 📖 Read the Medium article

• 05 Mar 2026 Korean Air's AI-Powered Drone Swarms: Multi-Agent Reinforcement Learning for MRO Inspection ✈️ We train a 5-drone swarm with tabular multi-agent Q-learning to cover an aircraft inspection grid for MRO, reducing collisions and wasted steps. 📖 Read the Medium article

• 28 Feb 2026 Route Vulnerability Clustering: Which Canada-U.S. Air Routes Break First? ✈️ We run K-Means and hierarchical clustering with UMAP on a small local Canada-U.S. route dataset, then overlay WestJet’s Summer 2026 cancellations to surface fragile route archetypes. 📖 Read the Medium article

• 18 Feb 2026 AI Incident Response Agent: SunExpress 737 Gear Collapse at Antalya ✈️ We implement a local AI agent using Ollama and LangChain to automate incident response checklists, draft communications, and log actions for a taxiway gear collapse scenario. 📖 Read the Medium article

• 12 Feb 2026 Baggage X-ray Anomaly Detection with Convolutional Autoencoders ✈️ We build a lightweight PyTorch convolutional autoencoder on synthetic baggage X-ray images to learn normal patterns and flag anomalies via reconstruction error—mirroring DXB’s AI-assisted hand-luggage screening. 📖 Read the Medium article

• 09 Feb 2026 Agentic AI for Airline Hub Disruption Management ✈️ We build a local multi-agent system (CrewAI + Ollama) that queries a read-only hub ops SQLite snapshot to draft a storm disruption action plan with crew duty-limit checks. 📖

Contributing

Contributions are welcome! Please feel free to submit issues or pull requests for improvements or new notebooks.

License

This repository is licensed under the MIT License. See the LICENSE file for details.