Airline Network Revenue Forecasting & Route Optimization
An integrated fleet and network optimization platform that uses predictive algorithms and large-scale network optimization to maximize airline revenue across route networks.
Challenge
Airlines operate complex route networks where small optimization improvements across hundreds of routes can yield massive revenue gains. Traditional planning methods couldn't keep pace with market dynamics.
Solution
Built a Spark-based big data platform with logistic regression forecasting models and large-scale network optimization algorithms.
Outcome
Deployed across two major Chinese carriers, enabling data-driven route planning decisions and measurable revenue improvements.
Intelligent Fleet Planning System
An AI-driven aircraft scheduling platform using operations research and large-scale integer programming to optimize fleet allocation and maintenance planning.
Challenge
Fleet planning requires balancing aircraft availability, maintenance schedules, route requirements, and regulatory constraints.
Solution
Developed a mathematical optimization engine using large-scale integer programming.
Outcome
Automated fleet planning, reducing planning cycles and improving aircraft utilization rates.
Revenue Management System
A comprehensive revenue management system implementing approximate dynamic programming and choice-based modeling for network-level seat inventory optimization.
Challenge
Revenue management requires real-time decisions on thousands of booking requests.
Solution
Implemented ADP for network optimization, Choice-based RM, EMSR calculations, and Bid Price Control.
Outcome
Enabled intelligent, network-wide revenue optimization with customer behavior-aware pricing.
Crew Scheduling & Rostering System
A regulatory-compliant crew scheduling system using integer programming and column generation for pilot and cabin crew assignments.
Challenge
Crew scheduling must satisfy CCAR-121, CCAR-135, FAR Part 117 regulatory requirements and union agreements.
Solution
Built an optimization engine using integer programming with column generation, integrated with ACARS.
Outcome
Improved average run-time by 30% with full regulatory compliance across Air China's network.
Irregular Operations Recovery Engine
A real-time dispatch and auto-recovery system using heuristic algorithms to rapidly reassign aircraft, crews, and passengers during disruptions.
Challenge
Disruptions require rapid recovery while minimizing passenger impact and operational costs.
Solution
Designed an OCC system with real-time dispatch, auto-recovery engine, and DCS integration.
Outcome
Significantly reduced recovery time during irregular operations for JetBlue Airways.
Passenger Accommodation System
Intelligent passenger rebooking with priority ranking, codeshare rebooking, and baggage tracking integration.
Challenge
Thousands of passengers need rebooking during disruptions accounting for class, loyalty, connections, and codeshare.
Solution
Built a system with priority ranking, codeshare rebooking, baggage tracking, and DCS/PSS integration.
Outcome
Automated rebooking, dramatically reducing gate agent workload and passenger wait times.
Dynamic Pricing & Fare Optimization
ML-powered dynamic pricing platform optimizing fare structures with ATPCO compliance and One Order standards.
Challenge
Thousands of fare products across hundreds of markets requiring real-time adjustment.
Solution
Developed an ML-based pricing engine with market segmentation and ATPCO integration.
Outcome
Enabled data-driven fare optimization across the network with intelligent market segmentation.
Air-Rail Intermodal Route Planning
Multi-modal route search engine with patented caching and route selection algorithms for high-concurrency performance.
Challenge
Massive route search space combining air and rail networks with high-concurrency requirements.
Solution
Designed with Dijkstra, A*, bidirectional BFS, 12306 API integration, and patented caching. 2 granted patents.
Outcome
Millions of route queries with sub-second response times, seamless air-rail intermodal booking.
Ancillary Product Recommendation Engine
Transformer-based recommendation system using collaborative filtering and ensemble models for personalized ancillary offers.
Challenge
Generic ancillary offers result in low conversion rates across diverse passenger segments.
Solution
Ensemble approach combining Collaborative Filtering, Transformer models, and XGBoost Ranking.
Outcome
Improved conversion rates through personalized, context-aware recommendations.
Intelligent Travel Assistant
AI-powered assistant combining LLM with AR navigation for natural language interaction and indoor wayfinding.
Challenge
Static information systems cannot handle diverse natural language queries or provide personalized guidance.
Solution
Agentic AI assistant with RAG, AR indoor navigation using WiFi, iBeacon, UWB, and RFID.
Outcome
Next-generation passenger experience platform with natural conversation and AR wayfinding.
Algorithmic Hedging Trading System
Low-latency execution engine for algorithmic hedging with dynamic hedge ratios and tail risk management.
Challenge
Ultra-fast execution with precise risk management under potentially illiquid conditions.
Solution
Low-latency engine with quote-driven strategy execution, spread calculations, and real-time visualization.
Outcome
Reduced quote latency from ~30ms to 8-10ms with precision risk controls.
MT5-Interactive Brokers Trading Bridge
Bidirectional trading bridge between MT5 and Interactive Brokers with order state machine synchronization.
Challenge
Bridging platforms with fundamentally different order models requires reliable bidirectional sync.
Solution
Order state machine for bidirectional sync, high-performance messaging, end-to-end order tracing.
Outcome
Seamless cross-platform trading with reliable synchronization and real-time risk monitoring.
MT5 Liquidity Gateway Middleware
Broker middleware aggregating liquidity from multiple providers with scalping detection and optimized pipelines.
Challenge
Aggregating multi-provider liquidity while detecting toxic flow with minimal latency.
Solution
Liquidity aggregator with MT5 Gateway API, scalping detection, and optimized order processing.
Outcome
~40% improvement in execution response time through pipeline optimization.
Shaila Community Platform
High-availability community platform for NFT enthusiasts with real-time recommendation and content moderation.
Challenge
Real-time content delivery, personalized recommendations, and moderation at scale.
Solution
Full-stack platform with real-time recommendations, UGC tools, and AI content moderation.
Outcome
Thriving community platform with effective content discovery and moderation.
LoopIn Food Delivery Platform
Comprehensive O2O food delivery platform with intelligent dispatch, dynamic routing, and real-time tracking.
Challenge
Real-time coordination between restaurants, drivers, and customers with peak hour management.
Solution
Full platform with geofence filtering, intelligent dispatch, dynamic routing, and fake review detection.
Outcome
Complete O2O platform with optimized delivery logistics and intelligent capacity management.
Aviation Standard File Parser
Specialized toolkit for parsing SSIM flight schedules and MCT minimum connecting time specifications.
Challenge
Aviation industry-specific data standards are complex to parse and integrate.
Solution
Built reliable parsing libraries for SSIM and MCT files with comprehensive test coverage.
Outcome
Streamlined data integration workflows for international business operations.