145 questions from 9 interviews · updated from reports 2015-2024
About
Juniper Networks makes networking products such as routers, switches, security systems, and software used by telecom, cloud, and enterprise networks. In India, it is commonly seen hiring software engineers, SDE-1s, and software development engineer interns for networking, systems, and cloud software work.
The roles that come up most are Software Engineer, SDE-1 and Software Development Engineer Intern. This covers 9 candidate interviews reported from 2015 to 2024. Most sat it at entry level (7 of 9 that recorded a level), with 2 internship interviews alongside. Among the 6 that recorded either route, arrivals split between campus drives (5, 83%) and off-campus applications (1, 17%). Most questions fall under CS fundamentals and DSA.
Ans. The OSI model describes networking in seven layers: physical, data link, network, transport, session, presentation and application. Each layer provides services to the one above it and hides lower-level details. In practice, TCP/IP is used more often: link, internet, transport and application layers, with IP handling addressing and TCP or UDP handling transport.
Ans. Dijkstra and Bellman-Ford find shortest paths from one source to all vertices in a weighted graph. Dijkstra repeatedly picks the nearest unvisited vertex, usually with a priority queue, and works only with non-negative edges. Bellman-Ford relaxes every edge repeatedly, handles negative weights, and can detect negative cycles. Dijkstra is faster; Bellman-Ford is more general.
Ans. MAC addresses identify network interfaces on a local link, IP addresses identify hosts across networks, ARP tables map IP addresses to MAC addresses, and routing tables choose where packets go next. The key distinction is scope: MAC and ARP work within a local network, while IP and routing work across interconnected networks.
Ans. Start with 0, 1, 1, then generate each next term as the sum of the previous three terms. Store the terms in a list if the full series is needed, or keep only three variables if printing. Run a loop until n terms are produced. Time complexity is O(n).
Ans. Inheritance is an object-oriented programming feature where one class derives from another class and reuses or extends its fields and methods. The derived class, often called a subclass, can add new behaviour or override existing behaviour, while the base class defines shared functionality. It supports code reuse and represents “is a” relationships.
Ans. Use two pointers, one for each sorted array, and repeatedly append the smaller current element to a new result array. When one array is exhausted, append the remaining elements from the other array. This keeps the merge sorted in one pass, using an output array, with O(n + m) time complexity.
Ans. A switch connects devices within the same local network, while a router connects different networks, such as a home LAN to the internet. A switch forwards frames using MAC addresses, usually at layer 2. A router forwards packets using IP addresses, usually at layer 3, and chooses paths between networks.
Ans. We use a switch with a router because they do different jobs: the switch connects devices inside the local network, while the router connects that network to other networks or the internet. The switch forwards frames using MAC addresses, and the router forwards packets using IP addresses, often also handling NAT and DHCP.
Ans. TCP flow control prevents a sender from overwhelming a receiver by limiting how much unacknowledged data can be in flight. The receiver advertises a window size in TCP acknowledgements, showing how much buffer space it has available. The sender must stay within this receive window, adjusting as the receiver drains or fills its buffer.
Ans. ARP maps an IP address to a MAC address on a local network, while DNS maps a human-readable domain name to an IP address. ARP uses local broadcast requests and caches replies. DNS uses a hierarchical naming system, typically queried through recursive resolvers, with cached results to reduce lookup time and traffic.
Ans. A PC boots by firmware starting the hardware, finding a bootable device, loading a bootloader, and handing control to the operating system kernel. The key detail is the firmware, usually UEFI or older BIOS, runs POST, initialises devices, reads the boot configuration, then the OS kernel starts drivers, services, and user space.
Ans. DHCP is a network protocol that automatically gives hosts IP configuration, such as an IP address, subnet mask, default gateway and DNS servers. DORA is the usual exchange: Discover from client, Offer from server, Request from client, and Acknowledgement from server. It commonly uses broadcasts and UDP ports 67 and 68.
Ans. Design it as a time-dependent graph service where airports are nodes and scheduled flights are edges with departure time, arrival time, airline, capacity and status. Ingest schedules and live disruption feeds into a durable store, index by origin and time, and use shortest path search with connection-time rules to return valid itineraries.
Ans. Use a balanced binary search tree storing key value pairs, such as a red black tree or AVL tree. get searches by key and returns the value in O(log n). get-next finds the in-order successor, keeping the smallest key greater than the target while descending. It also runs in O(log n).
Ans. VLANs are logical network segments created on switches to separate devices into different broadcast domains, even if they share the same physical network. They are used to isolate departments, improve security, reduce broadcast traffic, support guest networks, separate voice and data traffic, and make network management more flexible.
Ans. OSPF and RIP are interior routing protocols used to choose paths within an autonomous system. RIP is distance-vector, using hop count and periodic updates, with a 15-hop limit. OSPF is link-state, builds a full topology map, uses cost to run shortest path, and converges faster, especially in large networks.
Ans. A race condition happens when two or more concurrent operations access shared data and the final result depends on their timing or order of execution. It usually occurs when at least one operation writes to that data. The key prevention is to control access using synchronisation, such as locks, atomic operations, or transactions.
Ans. A compiler translates source code into executable code through a pipeline of analysis and generation stages. It lexes text into tokens, parses tokens into a syntax tree, checks types and semantics, optimises an intermediate representation, then generates target machine code or bytecode. The key idea is preserving program meaning while changing representation.
Ans. Implement htons by returning the 16-bit value unchanged on a big-endian host, and swapping its two bytes on a little-endian host. Network byte order is big-endian, so the key detail is detecting or assuming host endianness correctly. No data structure is needed, and the operation runs in constant time.
Ans. Use insertion sort by building a sorted linked list and inserting each original node into its correct position. Keep a dummy head for the sorted part, scan from it to find the insertion point, then relink pointers. This works well with linked lists because insertion is O(1) after search. Time is O(n²), space is O(1).
Ans. Merge sort recursively splits an array into two halves, sorts each half, then merges the sorted halves using a temporary array. Diagrammatically: [8,3,5,1] becomes [8,3] and [5,1], then [8],[3],[5],[1], then [3,8] and [1,5], finally [1,3,5,8]. It takes O(n log n) time and O(n) extra space.
Ans. Common types of software testing include unit, integration, system, acceptance, regression, performance, security, usability, smoke and exploratory testing. The most important distinction is between functional testing, which checks behaviour against requirements, and non-functional testing, which checks qualities such as speed, reliability, security and ease of use.
Ans. Test the table at its limits: zero rows, one row, maximum allowed rows, maximum plus one, zero columns, one column, maximum columns, and maximum plus one. Also test empty cells, minimum and maximum cell length, invalid data types, duplicate keys, very large values, special characters, sorting, filtering, and pagination at page boundaries.
Ans. A process is an independent running program with its own memory, while a thread is a smaller execution path inside a process that shares the process’s memory. Threads are cheaper to create and switch between, but need careful synchronisation. Multiprogramming means keeping several processes in memory so the CPU can switch when one waits.
Ans. OSPF and RIP are interior routing protocols that let routers learn paths inside an organisation’s network. RIP is distance-vector and chooses routes mainly by hop count, with a 15-hop limit. OSPF is link-state, builds a network map, and uses cost to find shortest paths, so it scales and converges better.
Ans. Reverse the 4-bit nibbles by repeatedly taking the lowest nibble and appending it to the result. For 0xaabbccdd, extract each nibble with mask 0xF, shift the result left by 4, OR the nibble in, then shift the input right by 4. This takes constant time and space for fixed-width integers.
Ans. Multithreading lets a program run multiple threads concurrently within one process, sharing memory but keeping separate call stacks. In C++, threads use std::thread, mutexes, locks, atomics and condition variables, with manual lifetime management. In Java, Thread, Runnable, executors, synchronized, volatile and concurrent utilities provide higher-level support managed by the JVM.
Ans. Socket-based client-server architecture uses network sockets as endpoints for communication between a client process and a server process. The server binds to an IP address and port, listens for connections, and accepts clients. The client connects to that address and port, then both exchange data using protocols such as TCP or UDP.
Ans. Convert the string to a character array, find the rightmost position where a character is smaller than the one after it, then swap it with the smallest larger character to its right and reverse the suffix. If no such position exists, it is already the last permutation. This takes O(n) time and O(n) space.
Ans. The browser resolves the URL to an IP address, opens a connection to the server, sends an HTTP request, receives a response, and renders the page. The key detail is DNS resolution happens first, followed by TCP and usually TLS setup, then HTML parsing, resource fetching, CSS layout, JavaScript execution, and painting to the screen.
Ans. An IPv4 header is a 20 to 60 byte structure containing version, header length, type of service, total length, identification, flags, fragment offset, TTL, protocol, header checksum, source address, destination address, and optional options. The most important detail is that it carries routing and fragmentation metadata, not application data.
Ans. Use merge sort on the doubly linked list, because it does not need random access and can relink nodes directly. Split the list with slow and fast pointers, recursively sort both halves, then merge them while fixing both next and prev pointers. The time complexity is O(n log n), with O(log n) recursion stack.
Ans. Network Address Translation lets devices on a private network share one or more public IP addresses by rewriting packet IP addresses, and often ports, at the router. The router keeps a translation table mapping internal addresses and ports to external ones, so return traffic can be forwarded back to the correct internal device.
Ans. Regular languages are the simplest class of formal languages, describable by regular expressions and recognisable by finite automata. Context-free grammars define languages using production rules where one non-terminal can be replaced by symbols. The key difference is that CFGs can express nested or recursive structure, such as balanced parentheses, which regular languages cannot.
Ans. The OSI layers are Physical, Data Link, Network, Transport, Session, Presentation and Application. Physical includes Ethernet PHY, DSL and fibre standards. Data Link includes Ethernet, Wi-Fi, PPP and ARP. Network includes IP and ICMP. Transport includes TCP and UDP. Session includes RPC and NetBIOS. Presentation includes TLS, SSL and MIME. Application includes HTTP, DNS, SMTP, FTP and SSH.
Ans. NAT lets hosts on a private LAN reach outside networks by rewriting their private source IPs to the router’s public IP. In a topology with several clients behind one gateway, the gateway also rewrites source ports and stores mappings in a NAT table, so returning packets are translated back to the correct internal host.
Ans. Track all allocations during startup and verify that each is freed or intentionally retained after initialisation completes. Use allocator instrumentation or tools such as Valgrind, ASan, kmemleak, or heap tracing to record allocation sites. Take memory snapshots at key boot phases and compare them, treating steadily growing unreachable allocations as leaks.
Ans. Common sorting algorithms trade simplicity, speed, memory, and stability. Bubble, selection, and insertion sort are simple but usually O(n²), with insertion good on nearly sorted data. Merge sort is stable O(n log n) but uses extra memory. Quick sort is usually fastest in practice O(n log n), but worst case O(n²). Heap sort is O(n log n) in-place but not stable.
Ans. Synchronization problems occur when multiple processes or threads access shared resources concurrently and the result depends on timing. Common issues include race conditions, deadlock, starvation and priority inversion. Operating systems use locks, semaphores, monitors and atomic operations to enforce mutual exclusion, ordering and safe communication between concurrent tasks.
Ans. SMTP is an application layer protocol used to send email between clients and mail servers, and between mail servers. It uses DNS MX records to find the recipient’s server, then exchanges commands like HELO, MAIL FROM, RCPT TO and DATA. Across OSI layers, SMTP sits on TCP, usually ports 25, 587 or 465, over IP and lower network layers.
Ans. Use a simple frequency array if the value range is known: scan the array, increment the count for each value, then print values whose count becomes 2. This avoids STL and reports each duplicate once. It takes O(n) time and O(k) extra space, where k is the value range.
Ans. Mutexes and semaphores are used to synchronise concurrent threads or processes so shared resources are accessed safely. They prevent race conditions and inconsistent data. A mutex gives exclusive access to one owner at a time, while a semaphore uses a counter to allow limited access or signal availability.
Ans. Use Floyd’s slow and fast pointer method to detect a loop. Move slow one node and fast two nodes each step; if they ever meet, a loop exists, and if fast reaches null, there is no loop. Diagram: A → B → C → D → E, with E pointing back to C.
Ans. Multithreading is running multiple threads within one process so work can happen concurrently while sharing the same memory. The main issues come from shared state: race conditions, deadlocks, starvation, livelocks, and visibility problems. These are managed with synchronisation, locks, atomic operations, thread-safe data structures, and careful design to minimise shared mutable data.
Ans. A race condition occurs when two or more processes or threads access shared data at the same time and the final result depends on the unpredictable order of execution. The key issue is unsynchronised access to a critical section, which can cause inconsistent data. It is prevented using locks, mutexes, semaphores, or atomic operations.
Ans. Devices include hubs, repeaters, bridges, switches, wireless access points and NICs at the network access layer, routers and layer 3 switches at the internet layer, firewalls and load balancers at the transport layer, and proxies, gateways and application firewalls at the application layer. The key point is their layer of operation.
Ans. Physical uses repeaters, hubs, modems and cables; data link uses switches, bridges, NICs and access points; network uses routers and Layer 3 switches; transport uses firewalls and load balancers; upper layers use gateways, proxies and application firewalls. The key point is that devices are classified by the layer whose header or signal they mainly inspect.
Ans. Use a stack, not a single regex, to check balanced parentheses because regex cannot reliably match arbitrary nesting. Scan characters left to right, push opening brackets, and pop only when the matching closing bracket appears. If a mismatch occurs or the stack is non-empty at the end, it is unbalanced. Time complexity is linear.
Ans. Paging is a memory management scheme that divides virtual memory into fixed size pages and physical memory into frames, letting the OS map pages to any free frames. It is required to avoid contiguous allocation, reduce fragmentation, and isolate processes. Virtual memory uses page tables and swaps missing pages from disk on demand.
Ans. Break the problem into given facts, what is being asked, and the formula or logic needed. For quantitative questions, convert units, set up equations, and calculate carefully. For logical reasoning, identify patterns, relationships, or assumptions. Eliminate impossible options first, then verify the remaining answer against the original question.
Ans. Scan from right to left to find the first character smaller than its next character, then swap it with the smallest character to its right that is larger, and reverse the suffix. This gives the next lexicographic permutation. If no such character exists, the string is already the largest permutation. Time complexity is O(n).
Ans. C is widely used in embedded systems because it gives very direct control over hardware with low runtime overhead. It maps well to memory registers, interrupts, and constrained RAM or flash, while producing small, predictable binaries. OOP is less important than determinism, portability across microcontrollers, and access to mature compilers and toolchains.
Ans. Messages are sent by breaking data into packets, adding addressing and control information, and passing them through network layers to the destination. Each packet carries source and destination addresses, such as IP addresses. Routers forward packets across networks, and the receiving device checks, orders, and reassembles them into the original message.
Ans. Represent present ages with variables, such as x and y. Translate each statement into an equation, keeping track of time: “after 5 years” means add 5 to each age, and “5 years ago” means subtract 5. Use ratios or differences as given, solve the equations, then check ages are realistic and non-negative.
Ans. Use alligation to compare the cheaper and dearer components with the required mean. Write the differences diagonally: dearer minus mean and mean minus cheaper. These differences give the ratio in which the two components must be mixed. For more than two components, first convert quantities to weighted averages, then apply the same idea.
Ans. Label one die 0, 1, 2, 3, 4, 5, 6, 7, 8. Label four faces of the other 0, 9, 18, 27, with the remaining faces repeats. Add the upward faces. Every n from 0 to 31 is 9q + r, with q from 0 to 3 and r from 0 to 8, so all values are covered.
Ans. I solve logic puzzles by defining the facts, listing assumptions, and reducing possibilities step by step. I would speak my reasoning aloud, check each conclusion against the original conditions, and revise if a contradiction appears. Without a specific puzzle, there is no single answer, but the method is structured elimination and verification.
Ans. Choose a real work situation with clear stakes, your specific responsibility, and a decision you had to make. Emphasise how you assessed options, involved others, managed risk, and acted under constraints. Interviewers listen for sound judgement, ownership, communication, learning from the outcome, and behaviour that fits their values and role expectations.
Ans. Choose a real, relevant workplace example with a clear decision, constraint, and outcome. Emphasise how you assessed options, involved others, managed risk, and acted professionally. Interviewers listen for sound judgement, accountability, communication, and learning. Use enough context to make the situation clear, but focus mainly on your actions and results.
Ans. Pick a realistic situation where you stayed calm, protected yourself, and used the right support channels. Emphasise not retaliating, keeping records, setting clear boundaries, speaking to a trusted teacher or counsellor, and prioritising safety. Interviewers listen for maturity, self-control, respect for rules, and willingness to seek help when needed.
No question matches those filters. Reset filters.
Showing 60 of 145 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 (63%) and DSA (28%).
Candidate interviews show an average of 4.6 rounds per experience, with a typical sequence of Online test → Technical → Technical → HR. Individual interview paths can vary.
Among questions with a recorded difficulty, the mix is easy 39%, medium 53%, hard 8%.
