Project Story

Inspiration

Flight cancellations in the United States routinely trigger massive operational losses. A single disruption can generate hotel costs, meal vouchers, compensation, and missed recovery flights simply because rebooking is slow and manual. The scale of waste — often reaching billions annually for major carriers — created the motivation to design an automated system that reacts faster than any human queue can. The goal was to eliminate avoidable overnight costs and transform recovery into an instant, data-driven process.

What We Learned

We learned how irregular operations cascade through an airline’s network, how recovery flights are selected, and how operational delays generate exponential financial damage. The project required deep understanding of:

• real-time aviation data flows
  
• constraints in airline rerouting logic
  
• practical rebooking rules
  
• API limitations and latency
  
• the economics of flight disruptions
  

We also explored how the same optimization logic applies to cargo logistics, where delays cause spoilage, SLA breaches, and refund losses. This showed that automated rerouting is not only viable but extremely scalable.

How We Built It

The system uses the OAG Flight Info v2 API as its backbone. The pipeline runs as follows:

1. Detect a cancelled flight event.
   
2. Pull all same-route flights using the live API.
   
3. Filter by departure time .
   
4. Score each candidate .
   
5. Select the earliest feasible flight.
   
6. If no same-day option exists, compute minimum hotel-window duration and assign next-day recovery.
   
7. Display a full rebooking plan for all passengers.

The frontend provides a clean table of cancelled flights, rebooked flights, hotel-required overflow cases, and multi-leg fallback routing. The logic is built to be API-driven, not hardcoded.

Challenges We Faced

• API rate limits and data coverage required careful request batching.
  
• Handling missing or inconsistent seat-availability fields demanded fallback rules.
  
• Time-zone conversion across airports created logic errors early in development.
  
• Designing a UI able to show complete rebooking chains without horizontal scrolling forced multiple layout iterations. (Still not the best UI design, but we did what we could in a short time)
• Creating a monetisation model that aligns with performance-based savings required research into airline economics.
  
• Ensuring the system can scale to cargo, produce, and parcel logistics introduced additional constraints around temperature windows, SLA cutoffs, and warehouse timings.

Despite these challenges, the final system demonstrates that cancellations can be turned from chaotic, expensive events into fast, automated recovery moments.

Comments about hosting - You'll need to locally host the website. We recommend Visual Studio Code. You will need 2 extensions. Live Server and Java. Upload the extracted file into the explorer and then right click on landingpage.html and Open with live server

We choose to code this the good ol way, fueled by caffeine and ChatGPT and a little bit of Claude.

Built With

• caffeine
• chatgpt
• claude
• elevenlabs
• oag