Taming Turbulence with Code: Developing AI-powered Solutions for Airline Disruptions
summary
An AI agent ingests real ops data and recommends schedule edits—mostly two-way swap actions—that the operation controllers can accept or reject; the system is Python/Java microservices with a React UI, uses event sourcing and reinforcement learning, and captures Ops Plus+ changes via divisum change data capture. Training replays historical event logs (a built “time machine”), runs near real time (≈20 seconds), started as a pre‑COVID POC and reached customers by 2021, team ~5, acceptance rates varying from ~40% down to ~10%.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:06Lanza Systems aviation software
- 2:55Flight operation challenges
- 7:14Schedule disruption AI
- 9:05User interface logic
- 12:20AI system internals
- 13:23Technology stack event sourcing
- 16:11Reinforcement learning agent
- 19:18AI recommendation rules
- 21:24AI architecture future
- 21:46Q&A
related talks
22mTaking Flight with AI and ML: Revolutionizing Airline Ops
22mMicrodosing the Economy
21mFrom Reactive to Proactive: How AI is Redefining Resilience in Modern Networks
47mHow agentic AI solves complex scheduling problems
18mRevolutionizing Operations with Intelligent Automation
13mAI Agents for Autonomous Industrial Operations on
59mBuilding resilient event-driven architectures
30mReal-Time ML services: From ideation to production