Mercor interview questions

Showing 1–30 of 44 results

• Behavioral: The Project You Are Leading and Why You Are Its OwnerMedium

  Describe the project you are currently leading, then defend why you are its owner rather than just a participant. The behavioral question tests whether you can explain a project crisply and back up a claim of ownership with specific decisions, risks you handled, accountability for setbacks and honest boundaries.

  MercorSoftware EngineerBehavioral & LeadershipTechnical Screen
• Build a Candidate Search SystemMedium

  Design an end-to-end talent search and matching engine that combines semantic vector retrieval with structured filtering to handle unstructured resumes, ambiguous queries, and strict hiring constraints.

  MercorMachine Learning EngineerML System DesignOnsite
• Build an Ollama Eval HarnessMedium

  This question evaluates a candidate's ability to design and implement a lightweight evaluation harness for a locally hosted language model, testing competencies in ML system design, service integration, metrics instrumentation, reliability engineering (timeouts, retries), concurrency control, configuration management, and extensibility.

  MercorMachine Learning EngineerML System DesignTechnical Screen
• Charge-Then-Save: Making a Checkout Handler Safe Under RetriesMedium

  Review a ticket-checkout handler that bills a card through a remote gateway and then saves an order row, explain every way it breaks when calls fail or are retried, and redesign it. Tests reasoning about partial failure, idempotency, and reconciliation.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Check Whether Every Parcel Finds a Locker With Forward OverflowMedium

  Parcels arrive one at a time for a bank of numbered lockers, each requesting a preferred locker and overflowing to the next higher-numbered empty one if it's taken. Decide whether every parcel finds a locker. Tests precise rule-following and an efficient alternative to simulating forward scans one slot at a time.

  MercorMachine Learning EngineerCoding & AlgorithmsOnsite
• Check-and-Insert for a Million User IDs: Balancing Latency, Memory and AccuracyMedium

  Design a check-and-insert service for about a million user IDs that reports whether an ID already exists and adds it if not. It tests memory estimates, data structure choice, atomic inserts under concurrency, trading memory for accuracy with compact filters, and durable on-disk storage with fast reads and writes.

  MercorSoftware EngineerSystem DesignTechnical Screen
• Code Review: A Ride Fare Method That Charges a Card Then Logs the TripMedium

  Review a Java method that charges a rider's card through an external processor and then writes a trip record, and explain what goes wrong if the caller fails mid-call and how to fix it.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Code Review: Find the Problems in a Python Dict-Based Memoization CacheEasy

  A Python code review question about a function that memoizes an expensive computation in a module-level dictionary. It tests whether a candidate can find correctness, memory, concurrency and key-equality problems in simple caching code and propose fixes with clear trade-offs.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Code Review: Improve a Java Method That Builds a CSV StringMedium

  A Java code-reading question: review a short method that concatenates a list of rows into a CSV string and propose improvements. It tests knowledge of string-building performance, separator and escaping rules for CSV, null handling, streaming output for large exports, and risks when files are opened in spreadsheets.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Code Review: Partial Failure in a Java Payment Method That Charges Then RecordsMedium

  A Java code-reading question about a payment method that charges an external provider and then inserts a database record, focused on what happens when the client fails in the middle of the call. It tests reasoning about partial failures, retries, idempotency keys, payment states, and reconciliation with the provider.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Code Review: Thread Safety of a Python Compute-and-Cache FunctionMedium

  Review a Python function that checks a dictionary cache, runs an expensive computation on a miss, and stores the result, assuming concurrent callers. It probes thread safety under simultaneous misses, making the computation run once per input, where to place a lock without serializing cache hits, and how Python dictionaries behave under concurrent reads and writes.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Compare Merge Sort with Factorial and Randomized Inefficient SortsHard

  Tests deep understanding of sorting algorithms, contrasting optimal divide-and-conquer strategies (Merge Sort and its adaptive variants) against factorial and randomized paradigms (Permutation Sort and Bogo Sort) across worst-case and expected complexity bounds.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Compare processes and threads in Linux and PythonMedium

  This question evaluates understanding of operating-system level concurrency versus language-level concurrency, specifically distinctions between processes and threads on Linux, the Global Interpreter Lock in Python, and trade-offs between Python threading and multiprocessing for performance.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Currency Exchange Rate Between Two Currencies via Chained ConversionsMedium

  Given a list of pairwise currency exchange rates such as USD to GBP at 0.77, compute the rate from a source currency to a target currency, converting through intermediate currencies and using rates in reverse where needed. It tests reasoning about chained conversions, unknown or unreachable currencies, and a clear time and space complexity analysis.

  MercorSoftware EngineerCoding & AlgorithmsOnsite
• Debug a Java CSV Export Function for Output and Performance BugsMedium

  A debugging exercise on a small Java function that converts rows of string fields into CSV text for import by another system. You find what is wrong with its output and why it slows down sharply on large exports, then fix it. It tests string handling, the cost of building strings in loops, and edge-case tests.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Design Candidate Search And LLM Evaluation TasksMedium

  Prepare for an MLE system design interview that combines candidate search, ranking, comparator reliability, and language-model fundamentals. The prompt covers retrieval, ranking feedback loops, top-k reasoning, transformer concepts, and decoding trade-offs without requiring proprietary hiring data.

  MercorMachine Learning EngineerML System DesignOnsite
• Design Work Orchestration for Machine-Learning Data PipelinesMedium

  Design a resilient, multi-tenant orchestration engine tailored for machine learning pipelines that manages versioned DAGs, heterogeneous compute routing, data quality gates, and atomic dataset publication.

  MercorMachine Learning EngineerML System DesignOnsite
• Design a DAG Pipeline Runner for Data-Labeling JobsMedium

  Design a multi-tenant service that runs user-defined DAGs of automated steps (pre-processing, model pre-labeling, QA checks, exports), tracks every run, and survives crashes. Tests data modeling, separating the scheduler from workers, durable state, retries and idempotency, tenant fairness and observability.

  MercorMachine Learning EngineerSystem DesignOnsite
• Design a Deliberately Inefficient Sorting Algorithm and Analyze Its ComplexityEasy

  An algorithms discussion question that asks for a deliberately inefficient but correct sorting algorithm and a careful analysis of its running time. It tests complexity reasoning about extremely fast-growing costs, worst-case versus expected time, and how such a sort compares with the lower bound for comparison sorting.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Design a High-Demand Event Ticket MarketplaceMedium

  Design an event ticketing platform capable of handling flash-sale spikes without overselling, featuring transactional seat holds, idempotent payment processing, and virtual queueing.

  MercorMachine Learning EngineerSystem DesignOnsite
• Design a Project-to-Contractor Matching SystemMedium

  Tests your ability to architect a two-sided matching platform that enforces strict eligibility filters, dynamic availability tracking, multi-stage ranking with exploration, and closed-loop model evaluation.

  MercorSoftware EngineerSystem DesignTechnical Screen
• Design a Social FeedEasy

  Design an end-to-end personalized social feed system capable of serving low-latency ranked content while handling hybrid fanout, multi-stage ML candidate retrieval, and stable pagination.

  MercorMachine Learning EngineerSystem DesignTechnical Screen
• Design a workflow orchestration system that runs multi-step jobs reliablyMedium

  A system design question asking you to build a workflow system where users define multi-step workflows with dependencies, start runs, and track progress. It tests data modeling, orchestration and worker architecture, durable state, retries and idempotency, fairness between tenants, scaling and observability.

  MercorMachine Learning EngineerSystem DesignOnsite
• Design an API to Deep-Copy a Dataset with Its Tasks, Model Responses and GradesMedium

  A system design question about an API that copies a dataset together with its tasks, model responses and grades so that the copy can be edited independently. It covers ID remapping, background jobs with progress tracking and resumable checkpoints, idempotent requests, worker leases, edits to the source during the copy, and cleanup after failure.

  MercorSoftware EngineerSystem DesignTechnical Screen
• Design the Worst Correct Sorting Algorithm and Analyze Its ComplexityMedium

  Asks you to propose the worst sorting algorithm that is still correct, explain how it works, argue that it always terminates with sorted output, and analyze its best-case and worst-case time and space. It tests counting arrangements, reasoning about factorial growth and stating what makes such a question well defined.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Evaluate Noisy Data for LLM Post-TrainingMedium

  Design a curation and validation framework to evaluate, filter, and safely integrate a large, noisy dataset into an LLM post-training pipeline without degrading model performance.

  MercorMachine Learning EngineerMachine LearningTechnical Screen
• Explain Kubernetes autoscaling and traffic routingMedium

  This question evaluates understanding of Kubernetes autoscaling mechanisms, external-to-pod traffic routing, and the differences between ReplicaSet and StatefulSet controllers.

  MercorSoftware EngineerSystem DesignTechnical Screen
• Explain what happens when opening a websiteMedium

  This question evaluates understanding of the end-to-end web request lifecycle, including DNS resolution, TCP/TLS handshakes, HTTP(S) exchanges, CDN and load balancer behavior, server-side handling, and the browser rendering pipeline.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Find the Probability That One Shop Supplies the Top Two CoffeesHard

  Calculate the probability that the top-ranked coffees in a blind competition come from the same shop under an exchangeable, tie-free ranking model.

  MercorSoftware EngineerStatistics & MathTechnical Screen
• Find the Production Bugs in a Dictionary Cache Around an Expensive ComputationMedium

  Review a short Python function that caches the results of an expensive computation in a dictionary, and identify what breaks in a long-running, concurrent production service. Tests reasoning about memory growth, eviction policy, concurrent cache misses, error handling, and input validation.

  MercorSoftware EngineerSoftware Engineering FundamentalsTechnical Screen