Oracle R&D PFE 2027 Technical Assessment: 10 Questions, Full-Screen Mode, and Preparation

The public Oracle R&D Morocco Center PFE 2027 HackerRank page lists 180 minutes and 10 questions, with an ML section labeled for ML applicants only. It also specifies full-screen mode. These are facts about that named assessment page, not a universal Oracle recruiting process. The official invitation landing page is the evidence for the counts and setup below.

For platform familiarization, use the sample test linked by the invitation.

The listed sections, without guessing the question format

Public section label	Listed questions	What remains unspecified
Problem Solving	1	Exact prompt and response format
Aptitude	3	Topic mix and permitted aids
Data Structures	2	Coding versus another question format
Algorithms	2	Difficulty and detailed syllabus
Machine Learning — ML applicants only	2	Applicant-specific inclusion and topic mix

The listed total includes the two conditional ML questions. Do not conclude that every non-ML applicant receives an eight-question test; the page does not establish that alternate configuration. The actual assigned test controls. It also says section order may change and the timer cannot pause.

There is no published cutoff or complete question bank in this evidence. An invitation page does not establish later interview rounds, shortlisting probability, or eligibility for every Oracle program. This guide deliberately stops at the visible assessment instructions.

Full-screen setup happens before pacing

The landing page says full screen activates when the test starts and instructs candidates to close browser sidebars, extensions, and developer consoles. Follow that setup and the sample-test flow before the timed attempt. A three-hour session also calls for power and stable connectivity; these are practical preparation steps, not extra Oracle eligibility rules.

Do not assume scratch paper, calculators, external IDEs, or unrestricted question revisiting are allowed. Look for those instructions in the actual test. HackerRank's during-test FAQ explains that section timers can prevent returning to earlier questions. That platform behavior is why a plan that always says “skip and come back” is unsafe.

Build a pacing plan from the interface you receive

Dividing 180 by ten gives 18 minutes per listed question, but that arithmetic is not a recommended equal allocation. A short aptitude item and a substantial implementation task can require very different effort.

What the actual test allows	Practical pacing choice
Free navigation with one shared timer	Inspect available questions, prioritize a tractable item, and reserve review time
Timed sections or restricted navigation	Finish checks before leaving a section; use its displayed budget
ML section assigned	Budget for it instead of assuming it is optional within your attempt
An item consumes time without progress	Move only if navigation permits; otherwise use remaining time for a simpler valid approach

This is a planning framework, not an Oracle scoring model. Before changing sections, distinguish running code from submitting the answer and check what the interface actually saved.

Preparation that matches the section labels

For data structures and algorithms, practice choosing a representation, stating complexity, and testing a counterexample. An implementation that passes one example can still mishandle duplicates, boundaries, or input size. Use the five records below to rehearse those specific failure modes.

For aptitude, practice interpreting a short quantitative or logical prompt accurately and checking units. The official page does not publish a syllabus; lists of guaranteed probability, sequence, or puzzle questions would overstate the evidence.

For an assigned ML section, rehearse explaining train/validation separation, leakage, and why a metric fits a stated objective. These are general ML preparation suggestions, not verified PFE question topics. If the invitation is unclear about inclusion, ask the assessment owner before starting.

Five focused coding exercises

Question	Focus	Why it belongs in this plan
Find the Earliest Pair with a Target Sum	hash maps, ordering, and linear-time reasoning	Use a timed first attempt, then explain the invariant and one failure case.
Validate Properly Nested Brackets	stack invariants and boundary cases	Use a timed first attempt, then explain the invariant and one failure case.
Find minimum subarray length with k distinct integers	sliding-window state and complexity	Use a timed first attempt, then explain the invariant and one failure case.
Group strings that are anagrams	canonical keys and grouping	Use a timed first attempt, then explain the invariant and one failure case.
Implement a Maximum-Tracking Stack	API design and duplicate values	Use a timed first attempt, then explain the invariant and one failure case.

If loading fails during the attempt

HackerRank's during-test FAQ recommends refreshing when questions or images fail to load and says progress and submitted answers are preserved. Logging out is different: it ends the attempt. If the error persists, report it through the test's support control and notify recruiting with the time, error, and remaining timer.

An interruption does not automatically pause the test or create an extension. For normal preparation, begin with Validate Properly Nested Brackets and explain your invariant before running the implementation.

HackerRank’s test-case documentation explains sample and hidden evaluation cases.

Sources and Further Reading

Documentation checked: September 6, 2026.

• official invitation landing page
• during-test FAQ
• HackerRank’s test-case documentation