139 questions from 19 interviews · updated from reports 2015-2024
About
Philips is a Dutch health technology company that makes medical imaging systems, patient monitoring equipment, and personal health products. In India, it is known for hiring Software Engineers, interns, and intern-to-full-time candidates for software, embedded systems, testing, and data roles.
The roles that come up most are Software Engineer, Intern and Intern + FTE. This covers 19 candidate interviews reported from 2015 to 2024. The largest group sat it at entry level (9 of 19 that recorded a level), with 8 internship interviews alongside. Among the 17 that recorded either route, arrivals split between campus drives (16, 94%) and off-campus applications (1, 6%). Most questions fall under CS fundamentals and DSA.
Ans. Break the problem into facts, rules, and conclusions. Translate each statement into simple conditions, then test what must be true, what could be true, and what cannot be true. Use tables, diagrams, or symbols if helpful. Eliminate answers that break a rule, and choose the option supported by all information.
Ans. String is immutable, while StringBuffer and StringBuilder are mutable classes for changing text. String creates a new object on each modification. StringBuffer is synchronised, so it is thread-safe but slower. StringBuilder is not synchronised, so it is faster and preferred for most single-threaded string construction.
Ans. SDLC is a structured process for planning, building, testing, deploying, and maintaining software. It gives teams a clear path from requirements to release, reducing risk and improving quality. Common stages include requirement analysis, design, implementation, testing, deployment, and maintenance, often repeated in agile or iterative models.
Ans. An abstract class is a base class that can share state and implemented behaviour, while an interface defines a contract that classes agree to implement. The key detail is inheritance: a class usually extends one abstract class, but can implement multiple interfaces, making interfaces better for capabilities across unrelated classes.
Ans. Compute the sum by iterating i from 1 to square root of n and adding divisors in pairs. If i divides n, add i, and add n / i if it is different from i and not equal to n. For n = 1, the sum is 0. Time complexity is O(sqrt n).
Ans. Cascade update is a referential integrity action where changes to a referenced key in a parent table are automatically applied to matching foreign keys in child tables. It is usually defined with ON UPDATE CASCADE. This keeps related records consistent and prevents broken references when a primary or unique key value changes.
Ans. A ROC curve shows how a binary classifier performs as its decision threshold changes. It plots true positive rate against false positive rate for many thresholds. A curve closer to the top left is better, and the AUC summarises this performance, with 1 meaning perfect separation and 0.5 meaning random guessing.
Ans. Identify what is being asked, list the given values, and choose the right formula or concept, such as percentage, ratio, time and work, speed, profit, or probability. Convert units if needed, form a simple equation, solve step by step, and check whether the answer is reasonable.
Ans. SQL joins combine rows from related tables, while transactions group database operations into one reliable unit of work. Common joins include inner join for matching rows, left join for all left-table rows, and right or full joins for broader inclusion. Transactions follow ACID principles, especially commit to save changes and rollback to undo failures.
Ans. Use two queues by making push costly: enqueue the new element into the empty helper queue, move all elements from the main queue into it, then swap the queue names. The front of the main queue is always the stack top. Push is O(n), while pop, top, and empty are O(1).
Ans. Identify the question type first, then write down the given facts clearly. Convert words into equations, tables, diagrams, or sequences where useful. Eliminate impossible options and check units, order, and conditions carefully. For reasoning puzzles, test one assumption at a time and verify the final answer against every statement.
Ans. Use an array or vector to store the factorial digits in reverse order and simulate multiplication digit by digit. Start with 1, multiply the stored number by each integer from 2 to n, carrying overflow across digits. This avoids fixed integer limits. Time complexity is roughly O(n × number of digits).
Ans. Use try-with-resources with BufferedReader to read and BufferedWriter or Files.write to write, so streams close automatically. Read each line into a String or process it immediately; use a List<String> only if all lines must be stored. Time complexity is O(n), where n is file size.
Ans. The prisoner should say, “You will shoot me.” If the policeman shoots him, the statement was true, so he should have been hanged instead. If the policeman hangs him, the statement was false, so he should have been shot instead. Both punishments break the rule, so he cannot be punished consistently.
Ans. IPC is inter-process communication, the ways separate processes exchange data or signals, and a race condition is a bug where the result depends on unpredictable timing between concurrent operations. IPC uses mechanisms such as pipes, sockets, shared memory or message queues. Race conditions are prevented with synchronisation such as locks, semaphores or atomic operations.
Ans. Partial functional dependency is when a non-key attribute depends on only part of a composite candidate key, rather than on the whole key. It matters in database normalisation because it causes redundancy and update anomalies. A relation with partial dependencies is not in Second Normal Form.
Ans. Semaphores are synchronisation primitives used by an operating system to control access to shared resources between processes or threads. A semaphore is usually an integer counter changed only by atomic wait and signal operations. Binary semaphores act like locks, while counting semaphores allow a fixed number of concurrent users.
Ans. A semaphore is a synchronisation primitive used by an operating system to control access to shared resources between threads or processes. It maintains a counter that is changed only by atomic wait and signal operations. A binary semaphore acts like a lock, while a counting semaphore allows a fixed number of concurrent users.
Ans. We use paging to let each process have a virtual address space while physical memory is managed in fixed-size blocks. Pages can be placed in any free frame, so memory need not be contiguous and external fragmentation is avoided. It also supports isolation, sharing, and moving pages between RAM and disk.
Ans. Read the question first, then read the passage or sentence carefully for meaning, tone and context. Identify keywords and eliminate options that are clearly wrong. For vocabulary, use surrounding words to infer meaning. For grammar, check subject-verb agreement, tense, articles, prepositions and sentence structure before choosing the best answer.
Ans. OOPS is a way to model software as objects that combine data and behaviour. A class is a blueprint, like Car, and an object is a specific car. Encapsulation hides engine details behind controls. Inheritance lets ElectricCar reuse Car features. Polymorphism lets different vehicles respond to start differently. Abstraction shows only essential details.
Ans. Number the jars 1 to n. Take 1 pill from jar 1, 2 from jar 2, and so on, then weigh the combined sample once. Compare with the expected weight if all were normal. The weight difference, divided by the contamination weight difference per pill, gives the contaminated jar number.
Ans. I would first profile the data, quantify missing values and noise, then choose cleaning rules based on the use case. Missing values can be imputed, flagged, or removed, while noisy values can be capped, smoothed, corrected, or treated as outliers. The key detail is to document assumptions and avoid introducing bias.
Ans. Cosine similarity measures how similar two vectors are by the cosine of the angle between them, while TF-IDF vectorization converts text into weighted term vectors. TF-IDF gives high weight to words frequent in one document but rare across the corpus, and cosine similarity then compares documents independent of their length.
Ans. The OSI model has seven layers: physical, data link, network, transport, session, presentation and application. Each layer abstracts part of communication, from bits on a wire to user-facing protocols. HTTP is an application-layer protocol using a request-response model, usually over TCP, or over TLS as HTTPS, to transfer web resources.
Ans. C++ uses stack allocation for local objects and heap allocation with new, usually released with delete or managed safely by RAII and smart pointers. Java allocates objects on the heap and deallocates them automatically using garbage collection when they are no longer reachable. The key difference is explicit lifetime control in C++ versus managed memory in Java.
Ans. Use an array when you need fast indexed access and mostly fixed size data, a linked list when you need frequent insertions or deletions in the middle, and a stack when access is last in, first out. The key trade off is access versus update cost: arrays index in constant time, linked lists do not.
Ans. Implement the stack with a singly linked list and keep a top reference to the first node. Each node stores the value and a next pointer. Push creates a new node and links it before top. Pop removes top and returns its value. Peek reads top. Push, pop and peek are O(1).
Ans. Start by identifying what is being asked, then write down the given values and the relevant formula or operation. Convert units if needed, simplify the arithmetic, and estimate the answer to check reasonableness. For speed questions, use distance equals speed times time. For basic maths, work step by step and avoid mental shortcuts that cause errors.
Ans. 22 times. The minute hand gains on the hour hand at a relative speed of 11 hours’ worth of angle every 12 hours, so it catches it 11 times in each 12-hour period. Over a full 24-hour day, that happens twice, giving 11 × 2 = 22 crossings.
Ans. Use the array itself as a marker: for each value x, look at index abs(x) minus one, and negate the value there if it is positive. If it is already negative, x is a duplicate. This works in O(n) time and O(1) extra space, assuming values are in 1 to n.
Ans. Turn switch A on for several minutes, then turn it off. Turn switch B on and enter the room. If the bulb is on, B controls it. If it is off but warm, A controls it. If it is off and cold, C controls it. Heat gives the extra observation needed.
Ans. Bubble Sort repeatedly traverses the linked list, compares adjacent nodes, and swaps them if they are in the wrong order. For a linked list, it is usually simpler to swap node data rather than relink nodes. Repeat passes until no swaps occur. It uses the existing linked list, runs in O(n²) time and O(1) extra space.
Ans. Operating system scheduling algorithms decide which ready process or thread runs on the CPU next, using policies such as First Come First Served, Shortest Job First, Round Robin, priority scheduling, and multilevel queues. Context switching is the mechanism that saves the current task’s CPU state and loads another’s, enabling multitasking but adding overhead.
Ans. Matrix factorisation using SVD decomposes a matrix A into three matrices: U, Σ, and Vᵀ, where U and V contain orthogonal singular vectors and Σ contains singular values. The most important use is low-rank approximation: keeping only the largest singular values captures the main structure while reducing noise, storage, and computation.
Ans. Read the question carefully and identify whether it tests grammar, meaning, tone, or inference. For grammar, check subject verb agreement, tense, articles, prepositions, and sentence structure. For vocabulary, use context clues and word roots. For comprehension, read the passage first, then match each option to evidence in the text.
Ans. About 15,000 two-wheelers and 4,000 four-wheelers. Assume a small town has 50,000 people, or 12,500 households at four people each. If 80% of households own one two-wheeler and 30% own a car, that gives 10,000 two-wheelers and 3,750 cars. Add commercial and extra vehicles, giving roughly these totals.
Ans. On a single-core CPU, the scheduler lets only one process run at a time and switches between the two by time-slicing or when one blocks. A timer interrupt or system call returns control to the kernel, which saves the current process state, chooses the next ready process, restores its state and resumes it.
Ans. Create two Runnable tasks, wrap each in a Thread, call start on both threads, then optionally call join to wait for both to finish. Use Thread objects as the core structure. The important detail is to call start, not run, because start creates concurrent execution. Time complexity is the combined work of both tasks.
Ans. In Java, local variables and method call frames are stored on the stack, while objects created with new are stored on the heap. Primitive local variables hold their values on the stack, but reference variables hold addresses to heap objects. Object fields live with the object on the heap, and heap memory is garbage collected.
Ans. Traverse once to the last node while keeping the previous node, then insert the new node between previous and last. If the list has zero or one node, insert the new node at the head. This uses only pointers, needs no extra data structure, and runs in O(n) time with O(1) space.
Ans. Use quicksort for general in-memory sorting, mergesort when stability or linked lists matter, heapsort when worst-case memory is tight, insertion sort for small or nearly sorted data, and counting or radix sort for bounded keys. Quicksort averages O(n log n) but worst O(n²); merge and heap are O(n log n); insertion is O(n²); counting is O(n + k).
Ans. Traverse the given directory recursively and return true when any encountered file matches the target path or name. Use depth first search with the call stack, or an explicit stack, to visit each child directory. The key detail is to handle permission errors and symbolic links safely. Time complexity is O(n), where n is entries visited.
Ans. Focus on core definitions, trace logic carefully, and explain trade-offs clearly. For data structures, know arrays, linked lists, stacks, queues, trees and hashing with time complexities. For C output, follow operator precedence and memory behaviour. For DBMS, revise keys, normalisation, joins and transactions. For OOP, know encapsulation, inheritance, polymorphism and abstraction.
Ans. OOP models software as objects that combine state and behaviour. Encapsulation hides internal data, abstraction exposes only useful operations, inheritance reuses and extends existing classes, and polymorphism lets different objects respond to the same method call differently. For example, Animal can define speak, while Dog and Cat override it. Method dispatch is typically constant time.
Ans. A trigger is database code that runs automatically when an event occurs, such as an insert, update, or delete; a stored procedure is a saved database program that is run explicitly by a user, application, or job. The key difference is control: triggers are event driven, while stored procedures are called deliberately.
Ans. Turn switch 1 on for several minutes, then turn it off. Turn switch 2 on and leave switch 3 off. Go to the bulb room. The lit bulb is controlled by switch 2. Of the two unlit bulbs, the warm one is controlled by switch 1, and the cold one by switch 3.
Ans. Label the dice 0, 1, 2, 3, 4, 5 and 0, 1, 2, 6, 7, 8, using 6 upside down as 9. Both dice need 0 for 01 to 09, and both need 1 and 2 for 11 and 22. This covers every date from 01 to 31.
Ans. Design a LibraryItem class with fields like title, id and availability, and methods such as borrow and returnItem. Book and DVD can inherit from it and override loanPeriod. A LibraryMember class borrows items through public methods, keeping data encapsulated. This shows abstraction, inheritance, polymorphism and encapsulation in a practical system.
Ans. About 1 to 2 lakh people. Start with population, say 20 lakh. Exclude small children and some non-tea drinkers, leaving about 12 lakh potential tea drinkers. Estimate what share drinks tea in the evening, perhaps 40 percent, and what share specifically at 4:30 PM, say 25 to 30 percent. That gives roughly 1.2 to 1.5 lakh.
Ans. Scan the encoded string left to right, reading each character and its following full number, and keep a running total of expanded length. When the running total becomes at least k, return that character. This avoids building the expanded string, uses constant extra space, and runs in O(n) time over the encoded input.
Ans. Create an HTML form with one text input named room_number and submit it to a backend endpoint such as /findStudent. The backend reads room_number, validates it, and runs a parameterised SQL query: select student_name from students where room_number = ?. Store data in a students table. With an index on room_number, lookup is efficient.
Ans. One egg is sufficient, if the aim is only to guarantee finding the threshold and there is no limit on drops. Start at floor 1 and move up one floor at a time. The first floor where the egg breaks is the threshold. This may take N drops, but uses only one egg.
Ans. The minimum worst-case number is 14 trials. Drop the first egg at floors 14, 27, 39, 50, and so on, reducing the gap by one each time. If it breaks, use the second egg to search linearly in the previous gap. Since 14 + 13 + ... + 1 = 105, 14 covers 100 floors.
Ans. Markov Chain Clustering finds graph clusters by simulating random walks, where flow becomes trapped in dense regions. Build a column-stochastic adjacency matrix M with self-loops. Iterate expansion M = M^e, usually e = 2, then inflation Mij = Mij^r / sum_k Mkj, usually r > 1. Prune small values and read clusters from the final matrix.
Ans. EM iteratively estimates hidden variables and parameters by alternating expectation and maximisation. For data X, latent Z and parameters θ, E-step computes Q(θ|θᵗ)=E_Z|X,θᵗ[log p(X,Z|θ)]. M-step updates θᵗ⁺¹=argmax_θ Q(θ|θᵗ). It increases or leaves unchanged the data likelihood p(X|θ) until convergence to a local optimum.
Ans. I would improve reliability, fairness, and user experience by adding a virtual waiting room, real-time seat locking, idempotent payments, better search, and clear booking status updates. The most important detail is preventing overselling: reserve selected seats with a short TTL, confirm them only after successful payment, and release them automatically on timeout.
Ans. Choose a situation where you stayed professional, addressed the behaviour early, and focused on the work rather than personalities. Emphasise listening first, setting clear boundaries, documenting serious issues, and escalating only when needed. Interviewers listen for maturity, discretion, conflict handling, and whether you protect team performance without creating more drama.
Ans. Pick two or three qualities linked to your best work, such as clear priorities, honest feedback, trust, and support when blockers appear. Ground them in a real example of a manager who helped you deliver. Emphasise how you respond to that style. Interviewers listen for self-awareness, maturity, and fit with their management culture.
Ans. Choose a real example where a junior was disengaged, underconfident, or struggling to grow. Emphasise understanding their goals, giving clear expectations, regular feedback, ownership, and recognition. Show you adapt your style to individuals. Interviewers listen for empathy, structure, accountability, and whether you motivate through support rather than pressure.
No question matches those filters. Reset filters.
Showing 60 of 139 questions. Ranked by how often the same question came back across interviews.
A spoken interview built from these questions, scored when you finish; the feedback is yours.
Start practisingWhen you are ready, record The One: a single interview hiring teams watch, so you stop repeating first rounds.
Candidate interviews most often cover CS fundamentals (52%) and DSA (26%).
Candidate interviews show an average of 3.4 rounds per experience, with a typical sequence of Online test → Technical → HR. Individual interview paths can vary.
Among questions with a recorded difficulty, the mix is easy 58%, medium 39%, hard 3%.
