Amadeus Nice, France
Software Development Engineer
Oct 2025 – Present
- Reduced pricing request latency from 67s to 1.8s using transaction-level memoization.
- Scaled supported pricing strategy rows per request from 10 to 100,000, enabling airlines to price large fare portfolios in a single request.
- Designed profile-based cache invalidation for airline availability, with estimated infrastructure savings of $80k/month across the first two customers, based on traffic, CPU usage, and cache hit rates.
- Rebuilt the invalidation pipeline with batching and exponential backoff, cutting a 1-hour workflow to 1 minute for critical airline operations.
- Restructured upstream request batching, increasing sustained throughput from 7 to 100 transactions per second.
- Driving design and implementation of an optimization across three distributed applications, resolving timeout and replication defects; projected savings of $50k/month.
- Led on-call operations for airline pricing and caching systems, resolving production incidents and data replication defects across inventory tiers.
- Eliminated redundant signals in Splunk production logs, reducing logging costs by 26%.