What problem is worth solving?
Airbus and Palantir launched Skywise in 2017, connecting work orders, components, fleet configuration, sensors and schedules. The platform extended from internal operations to airlines through analytics, applications and APIs, with early reliability, event-tracking and maintenance-support use cases.
Turn fragmented records into business objects teams can use to coordinate around an aircraft or component.
- WORKING ROLES
- Operations and maintenance specialists, data owners, platform and delivery engineers
How does an FDE turn symptoms into a solvable problem?
The following is an FDE exercise inspired by the case, not a claim about undocumented project events. Reported team actions appear in the context above.
Trace one interruption backward
Trace a synthetic equipment outage through asset IDs, component serials, maintenance and sensor timestamps. Identify missing relationships before choosing tables.
Look for semantic conflicts
Engineering defects, maintenance replacements and operational downtime are different objects and clocks. Compare how three roles describe the same event.
How does the solution enter the workflow?
- 01
Connect
Retain origin, keys, time and access scope for traceable inputs.
- 02
Model
Define asset, component, event and work-order relationships with temporal states.
- 03
Apply
Build a small affected-assets and open-work-orders view for expert review.
- 04
Productize
Expose reusable object queries and access controls; maintain domain rules separately.
Which decisions make the solution reliable?
Objects before dashboards
Require reliable relationships before broad dashboards; test join coverage and false merges.
Shared capabilities, scoped data
Reusable queries do not imply unrestricted data. Trace each field to its origin and visibility scope.
Who owns an alert after a component moves?
Synthetic scenario: component P-8 moves from asset A to B. Joining historical alerts using only its current owner corrupts history.
- Relationship
- Use valid_from / valid_to for installation relationships.
- Query
- Match each alert timestamp to the installation interval.
- Test
- Alerts before and after the move map differently; overlapping intervals are quarantined.
What should you do and deliver?
Inspect cross-source IDs and event time
- OUTPUT
- Data profile + join-quality report
- PASS CONDITION
- Fixtures reveal false merges
Model temporal relationships
- OUTPUT
- Ontology + object query
- PASS CONDITION
- Historical state is reconstructable
Add another application on the same objects
- OUTPUT
- API contract + transfer tests
- PASS CONDITION
- No duplicate private table logic
How do you prove completion and reuse the work?
Acceptance checks
Assess relationship quality, history, access isolation and effort for the second use case; validate operational impact in the workflow.
Reusable assets
Transfer the object, permission, lineage and action methods to retail or factories, while rebuilding the domain semantics.
Further reading and original sources
- Airbus launches Skywise ↗Airbus · 2017-06-19
- Ontology overview ↗Palantir · 查阅 / Accessed 2026-09-07