The Snowflake Online Assessment (OA)

The Snowflake online assessment, the OA, is a HackerRank or CodeSignal test that gates many software engineer and new-grad pipelines. It is commonly three algorithmic problems staged easy to hard in 90 to 120 minutes, any language allowed, scored on correctness and efficiency with every hidden test expected to pass. Dynamic programming, strings, graphs, and math dominate the set.

Treat it as a filter. The OA is graded automatically, often before a human reads your resume closely, and it is where a correct-looking solution that fails three hidden tests quietly ends the process. The problems read like standard mediums, but the assessment stacks them easy to hard and expects a full pass on each, so getting every edge case right matters as much as the idea.

Who gets an OA, and who skips it

The funnel opens with either an online assessment or a technical phone screen, depending on the team and the pipeline. New-grad and early-career pipelines are the ones most likely to hit a HackerRank OA; many experienced-hire loops go straight to a live CoderPad screen instead. Which one you get is set by the team, so ask your recruiter on the first call whether an OA is in your loop and which platform it runs on. If you are a student or new graduate, the new grad guide walks the lighter loop the OA gates.

What the OA contains

• Platform: HackerRank or CodeSignal. A timed, work-from-home test in the browser. Any language is allowed here, unlike the team-gated C++ round later in the loop.
• Problems: ~3, easy to hard. Commonly three algorithmic problems staged from easy to hard. The last one is where the difficulty lives, so budget your time so it does not get the leftovers.
• Time: 90 to 120 minutes. One sitting. Comfortable for a prepared candidate, tight for anyone rebuilding a pattern from scratch under the clock.
• Graded on: Full hidden-test passing. Correctness and efficiency, scored on every hidden test rather than partial credit. A solution that clears the sample but misses an edge case scores like a miss.

What the OA tests

The pattern mix leans toward dynamic programming, strings, graphs, and math. DP carries the assessment more than it carries the live rounds. It arrives with a correctness twist baked in: results taken modulo 1e9+7, intermediate sums that overflow a 32-bit integer, leading-zero and boundary cases, and constraints large enough that a brute force times out. A Premium question, job-scheduling DP, and ways-to-form-a-target-string are representative of the shape.

The reason the traps bite is the grading. Because the OA scores on full hidden-test passing, a correct recurrence that forgets the modulo wrap or overflows an accumulator fails silently on cases you never see. The fix is habit, not cleverness. Take the modulo at every step, reach for a 64-bit accumulator before it overflows, and walk your own boundary cases before you submit. The dynamic programming and correctness lesson drills exactly this rigor.

The Snowflake angle

The OA rewards the same instinct the live rounds do: correctness at scale before cleverness. Snowflake builds a system that moves petabytes, and its assessment reflects that. A clean solution that passes every hidden test beats a clever one that blows the time limit on the largest input. So prep to Hard for the OA even though a random tagged problem is a Medium. Spend your first hours where the loop spends its time: topological sort and dependency scheduling in graphs and topological sort, then the DP-with-correctness rigor above.

How to pass the Snowflake OA

Common questions