Skip to content

sergiulefter/skykit-optimizer

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

109 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SkyKit Optimizer

8th Place – Honorable Mention

HACKITALL 2025 – SAP 48H Section

SkyKit Optimizer is a full-stack airline logistics optimization system built in 48 hours during HACKITALL 2025 (SAP track).

Our solution ranked 8th overall, receiving an Honorable Mention for engineering quality, optimization strategy, and system design.


The Challenge

The SAP 48H competition simulated a real-world airline rotables logistics system.

Every hour (for 30 simulated days), the system had to:

  • Decide how many kits to load per flight
  • Decide how many kits to purchase
  • Avoid capacity overflow
  • Avoid leaving passengers without kits
  • Minimize total operational cost

Total score:

Transport + Processing + Purchasing + Penalties

Lower score = better performance.


Our Solution

SkyKit Optimizer is a stateful, adaptive optimization engine with real-time observability.

Instead of using fixed rules, the system:

  • Calibrates itself to the dataset
  • Learns from penalties during runtime
  • Dynamically adjusts buffers and thresholds
  • Balances overflow risk against stock starvation

Key Engineering Ideas

Dataset-Aware Calibration

Before simulation starts, the backend analyzes:

  • Network topology
  • Route distances
  • Demand distribution
  • Capacity limits
  • Penalty-to-cost ratio

From this, it computes dynamic purchasing thresholds and load factors.


Adaptive Penalty Learning

During the 720-round simulation:

  • Penalties are recorded in real time
  • Recurring issues adjust safety buffers
  • High-risk airports are detected
  • Economy loading factors are tuned dynamically

The strategy evolves during execution.


Capacity-Aware Purchasing

Purchasing decisions account for:

  • HUB capacity constraints
  • Current inventory levels
  • Spoke distribution
  • Forecasted demand
  • Overflow probability

This reduces:

  • Late-game overflow penalties
  • Over-purchasing
  • Capital lock-up

Structured Analytics

The backend exports structured penalty diagnostics:

  • Overflow grouped by airport and class
  • Unfulfilled passengers grouped by flight distance
  • Cost breakdown per penalty type
  • Comparable scoring metric (excluding identical end-game penalties)

Architecture

Backend (TypeScript / Node.js)

  • Simulation engine
  • Adaptive optimization module
  • Dataset calibration system
  • Penalty analytics engine
  • SAP evaluation API integration
  • Local API server for dashboard
  • Automatic eval platform lifecycle management

Frontend (React + TypeScript + Vite)

Real-time dashboard providing:

  • Live total cost tracking
  • Cumulative penalties
  • Comparable score
  • Inventory monitoring
  • Flight event timeline
  • Simulation controls

The frontend acts as the control center for the entire system.


Running the Project

You only need to start the frontend.

cd frontend
npm install
npm run dev

Open:

http://localhost:5173

From the dashboard:

  • Click Start Simulation
  • The frontend triggers the backend automatically
  • The evaluation platform starts automatically
  • The 720-round simulation runs
  • Results stream live into the dashboard

No manual backend startup required.


Why This Project Stands Out

This project demonstrates:

  • Algorithmic optimization under constraints
  • Adaptive runtime learning
  • Systems-level architecture design
  • Full-stack engineering
  • Real-time state management
  • Data-driven decision making
  • Clean modular TypeScript implementation
  • End-to-end automation (UI → backend → external platform)

All built within a strict 48-hour hackathon.


Result

  • 8th Place Overall
  • Honorable Mention
  • Delivered complete full-stack adaptive optimizer in 48 hours

License

Hackathon and portfolio project.
Built for HACKITALL I 2025 – SAP 48H Section.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages