PLTR-APD Palantir Application Developer – Associate
- Vendor
- Palantir
- Level
- Associate
- Questions
- 85
- Time
- 180 min
- Domains
- 8
- In the bank
- 485
Practice 485 questionsTake a timed mock examDay-by-day study plan
Guides for this exam
- Palantir Application Developer – Associate Study Guide: Build One Inspection App on Paper, Layer by Layer · 한국어
The exam at a glance
| Level | Associate |
|---|---|
| Questions | 85 |
| Time | 180 minutes (about 2.1 minutes per question) |
| Passing mark | 70% of the questions |
Domains and weights
The vendor does not publish domain weights for this exam, so the shares below are equal by assumption. Treat every domain as roughly the same size.
| Domain | Weight | Practice questions here |
|---|---|---|
| D1 · Ontology Schema Design and Workflow Assembly | 12.5% | 60 |
| D2 · Action Types and Write-Back Patterns | 12.5% | 61 |
| D3 · Ontology Functions | 12.5% | 61 |
| D4 · Workshop Application Development | 12.5% | 60 |
| D5 · OSDK, APIs, and External Applications | 12.5% | 62 |
| D6 · Automate and Operational Workflows | 12.5% | 61 |
| D7 · AIP Logic and AI-Powered Applications | 12.5% | 60 |
| D8 · Application Security, Governance, and Operations | 12.5% | 60 |
D1 · Ontology Schema Design and Workflow Assembly 12.5%
A business process or a schema change is described, and you choose how to model it in the Ontology or assemble it from Workshop, Action Types and Automate. Typical traps include reading cardinality the wrong way round, changing a live link without knowing its consumers, and storing team-specific labels as data. Good answers keep stored values and identifiers stable and place each step in the component built for it.
What to watch for
- Say the business rule out loud from one side: many tickets to one agent is many-to-one, and the foreign key sits on the many side.
- Before changing a link or object type, find out who depends on it. Workflow Lineage shows the consumers, and a branch gives room to migrate them.
- Split a workflow by role: Workshop for the user interaction, Actions for governed decisions, and Automate for checks, follow-ups and external calls.
Tasks in the official exam guide (4)
- 1.1 Object types, properties, links, interfaces, shared properties and value types
- 1.2 Stable identity, cardinality and meaningful relationships
- 1.3 Assembling end-to-end workflows across Ontology, Workshop, Functions, Actions and Automate
- 1.4 Workflow Lineage for impact, access troubleshooting and safe change
D2 · Action Types and Write-Back Patterns 12.5%
Questions about Action Types cover how changes are made, checked and secured. Scenarios include rules about who may act on what, concurrent users competing for the same record, external systems that need to hear about a change, and buttons that work for one user but not another. You pick the Action design or the first thing to check, and the answer usually depends on current state at execution time.
What to watch for
- Rules that depend on the target object or on who is acting belong in submission criteria checked at run time, not only in the form.
- When one user can act and another cannot, go through the layers one by one: Action permission, edit rights on the object, markings and criteria.
- If two edits must succeed or fail together, such as a check and a reservation, a function-backed Action is the place to make them atomic.
Tasks in the official exam guide (4)
D3 · Ontology Functions 12.5%
Functions are judged on efficiency, safety and maintenance, in situations with large linked datasets, bulk edits, a language the team no longer knows, and consumers that still run old logic after a publish. You choose how to query, test or version the function. Right answers push filtering and aggregation to the server and treat each published version as a contract; wrong ones load everything into memory or change things in place.
What to watch for
- Select the target objects with server-side filters or aggregations, including conditions on linked objects, rather than looping over everything in code.
- A bulk-editing function should be tried on a branch against the largest realistic input before it is published, then bounded or paged if needed.
- When a consumer still shows old behaviour, check which version it references. Consumers move to a new version only when someone updates them.
Tasks in the official exam guide (4)
D4 · Workshop Application Development 12.5%
Scenarios come from real Workshop modules: users who get lost, missing filter values, warnings that should appear, status changes, and an outside audience that needs a reduced view. The task is to pick the widget, variable, layout or deployment pattern that fixes the problem. Most correct answers connect widgets through variables and route every change through Actions instead of custom one-off logic.
What to watch for
- State changes in a module should go through an Action, so validation and permissions apply the same way as everywhere else.
- Conditional warnings and visibility come from derived variables. Shared variables also keep a selection in sync across pages and widgets.
- For different audiences, prefer a dedicated module built from shared parts with permissions on the data, over hiding pages in one module.
Tasks in the official exam guide (4)
D5 · OSDK, APIs, and External Applications 12.5%
External applications built on the OSDK and REST APIs are the topic. The scenarios ask when to build outside Workshop, which identity a call should use, why totals disagree between apps, and how to handle credentials and failures. Strong answers match authentication to whether a user is present, compute numbers on the server, and fix a specific cause instead of widening permissions.
What to watch for
- Ask whether a person is signed in. Per-user results call for user-based OAuth, while background jobs and partner servers use a narrowly scoped service client.
- Totals should come from a server-side aggregation with the same filters everywhere; summing pages in the client is where differences appear.
- Credentials and failures are part of the design. Scoped secrets that are rotated, plus error monitoring by type, help catch problems early.
Tasks in the official exam guide (4)
D6 · Automate and Operational Workflows 12.5%
Automate questions present an automation that misbehaves or is about to go live: double effects, failures nobody sees, the wrong trigger type, or a risky first run. You decide on triggers, conditions, ownership and rollout steps. The intended answers assume an event can arrive more than once and plan for safe testing and a quick stop, rather than hoping everything runs exactly once.
What to watch for
- Match the trigger to the need: events with conditions react to a specific change, and schedules suit regular summaries or cleanup.
- Make each effect safe to repeat. A flag, an existence check or a recorded reference stops duplicates when two events arrive close together.
- Ownership and rollout matter as much as logic. A team owner, team-routed alerts, test recipients and a staged start keep operations calm.
Tasks in the official exam guide (4)
D7 · AIP Logic and AI-Powered Applications 12.5%
AI features inside applications are tested on where AI fits, how much access it gets and how its work is checked. Scenarios include AI-backed Actions, prompt updates, summaries with broad access and Functions written by AI FDE. Correct choices keep the AI to drafting or interpretation with the smallest set of inputs and tools, and require evidence (evaluation, review, tests) before release.
What to watch for
- Give an AI feature only the objects and tools its task needs. A read-only task such as a summary should have no write Actions at all.
- Compare an old and a new prompt on the same evaluation suite with repeated runs, and look at results by category before releasing.
- Code from AI FDE gets the same review as human code, plus checks on schema, platform idioms, provenance and side effects such as retries.
Tasks in the official exam guide (4)
D8 · Application Security, Governance, and Operations 12.5%
Security and operations questions usually separate two things that look alike: opening an app versus reading its data, user-scoped execution versus a service identity, or a UI filter versus a real data control. You pick the control or signal that fits the exact need. Good answers change only the failing control, enforce rules at the data level, and give operations leads health signals per component.
What to watch for
- When an app opens but shows no data, the problem is data access, not app access. Check the object types and backing data for that user.
- Row or property limits for partners or other groups belong in data-level controls, because a filter in one module protects only that module.
- Know whose permissions a run uses: the invoker's in user-scoped mode, or a configured least-privilege service user otherwise.
Tasks in the official exam guide (4)
A study order that works
The 17-day study plan for PLTR-APD takes the domains in this order. Each day has a short reading, key terms and a quiz from the bank.
- Days 1–2: D1 · Ontology Schema Design and Workflow Assembly
- Days 3–4: D2 · Action Types and Write-Back Patterns
- Days 5–6: D3 · Ontology Functions
- Days 7–8: D4 · Workshop Application Development
- Days 9–10: D5 · OSDK, APIs, and External Applications
- Days 11–12: D6 · Automate and Operational Workflows
- Days 13–14: D7 · AIP Logic and AI-Powered Applications
- Days 15–16: D8 · Application Security, Governance, and Operations
- Day 17: Final Review: Exam Strategy and Cross-Domain Judgement
Recent changes to these questions
- : 4 answer choice sets changed, 4 answer keys corrected, 4 explanations rewritten
- : 60 questions added
- : 1 answer choice set changed, 1 answer key corrected, 1 explanation rewritten
- : 25 questions added
About these questions
Every question is original, written from the public exam guide and checked against the vendor's own documentation. None comes from a real exam. Each one has an explanation of why the answer is right and why the other options are wrong. How the questions are written.