What is an online assessment (OA) in a tech interview?

An online assessment (OA) is a timed coding test administered before any live interview rounds. It typically takes place on platforms like HackerRank, CodeSignal, or a company's proprietary system. Most OAs consist of 2-3 algorithmic coding problems to be solved within 70-120 minutes. The OA is a screening step: candidates who pass move on to phone screens or onsite interviews. Amazon, Google, Microsoft, and most major tech companies use OAs for new grad and internship hiring.

How hard are online assessment questions compared to LeetCode?

OA difficulty varies significantly by company. Amazon's OA for new grads typically contains medium-difficulty problems similar to LeetCode medium, with a focus on arrays, strings, and simulation. Google's CodeSignal OA (used for university recruiting) targets medium to medium-hard problems. Jane Street and Citadel OAs can include hard problems for quant and systems roles. In general, passing an OA requires comfortably solving LeetCode medium problems within a time limit.

What topics appear most in online assessments?

Based on real OA reports in the LeakCode database, the most common topics are: arrays and strings (always present), simulation problems (implement a process step by step), hash maps and sets for counting and deduplication, basic tree and graph traversal, sorting and binary search, and sliding window or two pointer patterns. DP appears in OAs but less frequently than in onsites. MCQ-style OAs at some companies also test data structures and time complexity knowledge.

How should I manage time during an online assessment?

Allocate time proportionally: for a 90-minute OA with 2 problems, spend no more than 40 minutes on the first problem before moving to the second. A partial solution with most test cases passing is better than spending the full time on one problem and leaving the second blank. Always submit a brute-force solution first to capture partial credit, then optimize if time allows. Read all problems before starting so you can pick the easiest one first.

How does LeakCode help with OA preparation?

LeakCode has thousands of real OA questions reported by candidates across Amazon, Google, Microsoft, and other companies. These are actual problems from real assessments, filtered for quality. Search for a company's OA questions on LeakCode to see which problem types and categories that company has used recently, so you can practice the right patterns rather than studying randomly.

OA Online Assessment Questions: Complete Guide (2026)

What online assessments test, which topics appear most, how to manage time effectively, and how to use real OA data from LeakCode to prepare for each company's specific format.

What is an online assessment and why it matters

An online assessment (OA) is the first real coding gate in most tech company hiring pipelines. It comes before any live interview and is used to screen candidates at scale. Companies like Amazon, Google, Microsoft, Citadel, and Goldman Sachs all use OAs for new grad, internship, and some experienced roles.

Unlike a live phone screen, an OA has no interviewer to ask clarifying questions. You work alone against a timer, submitting code that is automatically judged against hidden test cases. The pressure is different from live interviews: there is no partial credit for explaining your approach, only for passing test cases.

LeakCode has thousands of real OA questions across its 51,000+ entry database. These come from candidate reports on 1Point3Acres, Reddit, Blind, and LeetCode Discuss, classified by round type as "OA" or "Online Assessment." Knowing what a specific company's OA actually contains is a significant preparation advantage.

OA format by company type

Large tech companies (FAANG and equivalent)

2-3 algorithmic coding problems in 70-120 minutes. Primarily medium-difficulty array, string, and graph problems. Amazon's OA is one of the most well-documented: real-world-framing prompts that reduce to standard algorithmic problems. Google uses CodeSignal for university recruiting with standardized problem sets. Meta sends OAs for some roles via custom platforms.

Financial and quant firms

Jane Street, Citadel, Two Sigma, and similar firms use harder OAs targeting algorithmic thinking and sometimes math or probability. These OAs can include hard graph problems, game theory, or combinatorics that go beyond typical FAANG difficulty. Often 2-4 problems in 90-150 minutes.

Mid-size tech companies

Stripe, Airbnb, Lyft, and similar companies often use a "Karat" or custom interview format. Some use the CodeSignal general coding assessment which has a fixed 70-minute window with 4 problems. The scoring is percentile-based across all test-takers, and companies set their own passing thresholds.

Internship and new grad pipelines

OAs are most common in internship and new grad hiring, where companies need to screen hundreds or thousands of applicants cost-effectively. The difficulty is typically lower than experienced hire onsites, targeting LeetCode easy-to-medium range. Some companies include MCQ sections on data structures, complexity, or debugging.

Topics that appear most in OAs

Based on real OA reports across thousands of entries in the LeakCode database, here are the topics ranked by frequency:

1.Arrays and strings: manipulation, parsing, sliding window. Present in virtually every OA.
2.Simulation: implement a process step by step according to rules. Amazon uses this format heavily to create real-world-looking problems.
3.Hash maps and counting: frequency counting, deduplication, grouping. Nearly universal.
4.Basic graph traversal: BFS or DFS on grids or small graphs. More common in experienced hire OAs.
5.Sorting and binary search: custom sort comparators, binary search on answer.
6.Dynamic programming: appears in harder OAs, usually Fibonacci-style or 1D DP.

OA strategy: how to maximize your score

OAs are scored on test cases passed, not code quality or explanation. Here is how to maximize your score:

1.Read all problems first. Take 5 minutes to scan all problems before writing any code. Start with the one you are most confident about, not necessarily the first one.
2.Submit a brute force early. Get a working submission in the first 15-20 minutes, even if it is slow. This locks in partial credit. Optimize afterward if time allows.
3.Handle edge cases explicitly. Empty input, single element, all same values, very large or very small numbers. OA test cases often include edge cases that catch untested code.
4.Set time limits per problem. Allocate time before you start. For a 90-minute 2-problem OA: 35 minutes on problem 1, 35 minutes on problem 2, 20 minutes to debug and optimize. Stick to this.
5.Use your strongest language. OAs allow any language. Use whatever you are most fluent in, not what you think the company uses. Speed of implementation matters more than language choice.

Top companies with online assessments

Amazon Google Microsoft Meta Bloomberg Goldman Sachs Citadel JP Morgan

Find Real OA Questions on LeakCode

LeakCode indexes thousands of real OA questions from Amazon, Google, Microsoft, and more. Search by company to see what their assessment actually contains.

Browse OA Questions

How LeakCode helps with OA prep

LeakCode indexes real OA questions from 1Point3Acres, Blind, Reddit, and LeetCode Discuss, classified by round type. Filter by company and select the "OA" or "Online Assessment" round type to see exactly what a company's assessment has included in recent months.

OA Strategy: Read All Before Solving Any

Online assessments typically present 2-4 problems with a fixed time budget (often 60-90 minutes total). The winning strategy is to read all problems first, identify the easiest, and lock in that score before tackling harder ones. Many OAs use partial credit per test case, so a brute-force submission that passes 60-70% of tests can still advance you. Reports on LeakCode show that candidates who solve 2 problems partially often advance over candidates who solve 1 problem perfectly but never attempt the second.

Amazon OAs in 2025-2026 cluster around two-problem format with one medium array problem and one medium graph or DP problem. The advance threshold is roughly 70%; missing one easy edge case often drops you below it. Microsoft OAs typically include a debugging round where you fix existing code rather than write new code. Google OAs are less common but when used, they skew toward 3 problems with one hard sprinkled in.

Common OA Patterns by Company

Amazon OAs heavily favor arrays and strings with sliding window or two-pointer patterns. Practice the canonical 20 patterns on LeetCode tagged Amazon, sorted by recency. Microsoft OAs lean toward simulation problems and recursion-heavy questions; the Codility platform they use rewards clean, readable code over micro-optimization. Bloomberg OAs use HackerRank with focus on OOP design problems (implement an event system, a calendar, a basic LRU cache).

Reports on LeakCode show candidates who target a specific company's OA pattern in their final week of prep substantially outperform candidates who do generic LeetCode practice. Use the company filter to see only the most-recent OA reports for your target.