naquuuu@systems traveloka
traveloka logoflight products & operations // traveloka

flight products, refund accuracy & operations automation

Product Management • April 2024 - March 2026 • Flight Products & Post-Booking Automation

outcome first

the result this page leads with: flight refund calculation accuracy moved from 87.97% to 94.24%. the work behind it: direct provider api integration, plus a tax (vat, service charges) and commission calculation engine. refund closing automation doubled from 16.76% to 34.38%. refund policy completeness reached 95.76%.

the operational reality

managing airline ticketing and high-volume refund operations requires balancing commercial flexibility, complex carrier tariff structures, and strict regulatory standards. at traveloka, i owned commercial ticket development, led direct provider api integrations and tax calculation engines, and implemented automation to reduce operational workload while ensuring compliance with indonesian ministry of transportation pm 185 tahun 2015.

system flow [ dual-source convergence & anomaly feedback loop ]
refund policy completeness // dual-source convergencesimplified
provider apis, cat 16 fare rules & anomaly loop
traveloka refund policy completeness flow blueprint convergence flowchart: two upstream sources (direct provider apis and airline cat 16 fare rules processed by an improved llm-assisted parsing and translation workflow) converge into refund policy coverage, guarded by an anomaly dashboard that flags policy errors and loops back to coordinate provider fixes, resulting in a single measured output of 95.76% refund policy completeness. upstream provider apis carrier policy & fare data integration provider api integration structured policy ingestion audit anomaly dashboard flags policy errors cat 16 airline fare rules unstructured tariff text llm-assisted improved rule parsing translates cat 16 for policy policy interpretation merged api & cat 16 rules 95.76% completeness refund policy coverage [coordinate provider fix] lines flow dependency verified outcome failure or risk path shapes data source document action start / end
system architecture 01 [ commercial product & revenue logic ]
modified inventory ticket product
owned end-to-end development of a modified inventory ticket product, redesigning marked-up refund fee policies and discounted pricing logic. maintained strict compliance with indonesian ministry of transportation pm 185 tahun 2015.
commercial outcome: ~idr 10bn net revenue
generated approximately idr 10 billion in net revenue within 3 months of launch from a modified inventory ticket product with redesigned refund-fee policies and discounted pricing logic.
system architecture 02 [ calculation accuracy & provider api integration ]
tax & commission calculation engine
led direct provider api integration and engineered a calculation engine supporting new tax components (vat, service charges) and commissions, reducing customer and operational escalations.
accuracy improvement: 87.97% to 94.24%
improved flight refund calculation accuracy from 87.97% to 94.24% by leading direct provider api integration and engineering a tax (vat, service charges) and commission calculation engine.
system architecture 03 [ operations automation & fare rule interpretation ]
rpa refund closing automation: 16.76% to 34.38%
doubled refund closing automation from 16.76% to 34.38% through robotic process automation (rpa), cutting manual handling time and freeing operations capacity.
policy completeness (95.76%) & cat 16 fare rules
drove refund policy completeness to 95.76% via direct provider api integration and built an anomaly detection dashboard to flag illogical policy data for provider escalation. improved an llm-powered workflow to parse and translate airline cat 16 fare rules in real-time, designed to automate refund policy interpretation at scale.
failure modes & edge cases [ provider feeds and fare rules ]
illogical policy data
provider policy feeds can carry illogical policy data, so an anomaly detection dashboard flags policy errors for provider escalation instead of trusting the raw feed.
unstructured fare rules
airline cat 16 fare rules arrive as unstructured tariff text, so an llm-powered workflow was improved to parse and translate them in real time, designed to automate refund policy interpretation at scale.
trade-offs [ commercial flexibility against regulation ]
pricing freedom inside a regulated frame
the work sits between commercial flexibility, complex carrier tariff structures, and strict regulatory standards: refund fee and discounted pricing logic was redesigned while maintaining compliance with indonesian ministry of transportation pm 185 tahun 2015.
← view all work experience
open to collaboration

let's build something thoughtful together.

i'm open to full-time product roles, freelance projects, and collaborations. whether it's untangling complex software workflows or hands-on industrial design, i care about working with teams who value craft and solving real problems.

gmail ↗ whatsapp ↗ linkedin ↗