Articles

Filtered (Feenyx) Assessment: Prepare the Editor and Explain Your Solution

Prepare for a Filtered or Feenyx assessment with editor checks, an original coding exercise, a solution explanation, and a checklist before question submission.

A Filtered assessment invitation can create two different kinds of uncertainty: what the platform expects you to do, and whether your solution is ready to submit. Prepare for both. Rehearse the required editor, work from the actual input contract, and practice explaining why the code is correct before committing a question.

The central idea: Your preparation should reproduce the working conditions and the reasoning you need to demonstrate. A polished solution in a familiar local IDE is less useful if you cannot build, test, and explain it in the assessment environment.

Evidence boundary: Official Feenyx help pages, its website, and its extension listing were checked on October 4, 2026. This is a provider-workflow guide, not an employer-specific hiring report. We do not have two independent same-cycle candidate reports supporting a fixed question mix, duration, or cutoff. The coding exercise below is original practice, not a leaked Filtered or Feenyx question.

For a concrete rehearsal, open Debug a Python Aggregation Loop with Dictionaries and Lists. Try explaining the data structure and one failing case before changing the code.

Why do Filtered searches lead to Feenyx?

Observed official evidence: At the time of review, www.filtered.ai redirected to the Feenyx website, whose footer names Filtered, Inc. The Feenyx Interview Tool listing also identifies Filtered Inc. as its developer. These signals explain the two names in this title; they do not establish that every older invitation has an identical interface.

Use the invitation you actually received as the entry point. Verify the hiring company, role, deadline, question configuration, and permitted languages. A generic practice page or an older screenshot should not override current instructions for your assessment.

The current candidate guide includes separate instructions for basic coding, front-end, database, data-science, IDE, and other question formats. Avoid preparing as if every invitation were a single algorithm test. Conversely, the existence of those formats does not mean your interview includes all of them.

Prepare the environment before starting a timed question

Official guidance: Feenyx's candidate checklist names current Chrome or Firefox, asks candidates to confirm language support, and calls for working camera and microphone hardware. Read it alongside the invitation rather than assuming your usual browser or language version is available.

The extension requirement has a configuration boundary. Official fact: The dedicated extension explanation says hiring companies can toggle the requirement. That is more specific than older checklist wording that treats installation as universal. Follow your assessment's actual requirement and raise concerns with the hiring team before starting.

The extension listing describes access to open-tab titles and URLs associated with the interview and recommends closing other tabs. Treat that as the provider's stated behavior, not an independently audited privacy guarantee. Read the installation prompt and permissions yourself. An AI-branded hiring product does not, by itself, authorize candidates to use AI assistance.

Official guidance: The hardware troubleshooting page explains camera, microphone, sound-output, and network checks. For microphone detection, it suggests speaking or making a sound when a selected microphone is not registering. Confirm that the selected input is the one you intend to use.

Preparation recommendation: Check typing, indentation, keyboard shortcuts, and the run workflow in a sample environment. Find out where output appears and how a syntax error is presented. These small checks prevent an editor problem from consuming the same attention you need for algorithm design.

A useful readiness note records: browser tested, language confirmed, hardware check passed, question formats reviewed, deadline understood, and assistance rules clarified. Keep it brief enough to inspect before the actual start action.

Distinguish the code pad, IDE questions, and demo mode

Official fact: The interview walkthrough describes sample questions that resemble the working environment but have demonstration content and no submit button. It also distinguishes the overall time estimate from time limits that may apply to individual coding challenges.

Use a demo to learn controls and constraints. It cannot prove which hidden tests, scoring thresholds, or exact tasks your hiring assessment uses. A missing submit button in a sample is not evidence that the real assessment saves itself into a final submission.

Official fact: Feenyx's interview FAQ says code cannot be pasted from an external IDE into its coding interface and recommends writing the solution directly there. Rehearse under that condition rather than relying on a final copy-and-paste transfer.

Question type matters. The separate IDE-question guide describes a supplied VS Code environment, a terminal, and testing the solution during development. A basic code pad and a hosted project environment have different workflows. Determine which one you have before planning package installation or multi-file work.

For SQL, the database-question guide documents SQLite in its published environment description. That is a reason to check the actual SQL dialect rather than assuming production PostgreSQL or MySQL syntax. Treat published version tables as a starting reference; inspect the configured environment before depending on a version-specific feature.

Rehearse an original task with an explicit contract

Here is an original practice task that fits a small editor rehearsal: choose the latest event for each device.

The contract is deliberately precise. Each valid event is a three-element tuple: a nonempty string device ID, a nonnegative integer timestamp, and a value. Boolean timestamps are invalid in this exercise. Ignore malformed rows. Return a mapping from device ID to the entire selected event. When timestamps tie, the event appearing later in the input wins. Output ordering is not part of the contract.

Input, in order:
("a", 100, 3)
("b", 90, 2)
("a", 95, 9)
("a", 100, 7)
("b", None, 8)   -> invalid timestamp

Selected events:
a -> ("a", 100, 7)
b -> ("b", 90, 2)

Before coding, identify the invariant: after processing each prefix of the input, the mapping contains the selected valid event for every device encountered in that prefix. The selection rule compares timestamps first and input position second.

A dictionary supports one entry per device. For a valid new event, insert it if the device is absent. Otherwise, replace the stored event when the new timestamp is greater than or equal to the stored timestamp. The equality case implements “later input wins.” A strict greater-than comparison would silently implement a different tie rule.

This reasoning also explains why sorting is unnecessary for the stated contract. The scan uses expected O(n) time with dictionary operations and O(d) additional storage for d distinct devices. If the interviewer changes the output to a sorted list, include that additional sorting cost rather than carrying over the original complexity claim.

Original test	Expected behavior	Defect it exposes
Empty input	Empty mapping	Incorrect initialization or assumptions about a first row
Same device, newer then older event	Keep the newer event	Replacing unconditionally on arrival
Same device, equal timestamps	Keep the later input event	Using the wrong tie comparison
Invalid timestamp after a valid event	Preserve the valid winner	Comparing missing or invalid values
Two devices with different timelines	Select independently	Accidentally maintaining one global winner

These are contract tests, not a prediction of the platform's hidden tests. The value field is deliberately unconstrained; do not reject a zero value merely because it is falsey. If you choose a different validation policy, state it explicitly and keep your examples consistent.

Official fact: Feenyx's basic-coding guide distinguishes visible sample tests from hidden scoring tests. Passing a sample establishes compatibility with that sample. It does not establish that an implementation handles every promised input.

Explain the solution through a trace, not a code reading

Official fact: The candidate-experience guide describes a video explanation after submission as an optional setting. The basic-coding instructions also mention possible video or file explanations. Check whether your configured question requests one.

Use this original explanation framework for the event task:

Contract: “I select one valid event per device, and equal timestamps favor later input.”
Invariant: “After each processed row, the mapping holds the correct winner for the processed prefix.”
Trace: “The older a:95 event does not replace a:100; the later tied a:100 event does.”
Cost and risk: “The scan is expected linear time with one stored event per device. I still need to check malformed input and tie cases.”

This is a rehearsal structure, not a mandated platform script. Fit the actual response time and prompt. If the explanation asks for trade-offs, discuss the alternatives that change this task: sorting before selection, keeping only values instead of entire events, or rejecting malformed rows instead of skipping them.

If the solution is partial, describe the precise supported behavior and a counterexample. “The normal case works, but equal timestamps currently keep the earlier row” gives a reviewer useful evidence. Claiming correctness without testing the tie contract hides the most important limitation.

For a video response, the video-question guide documents different recording flows depending on whether re-recording is allowed. Check the displayed rule before beginning. Do not assume every explanation permits another take.

Make the submission decision before pressing Submit

Official fact: The interview walkthrough says submitted questions cannot be revisited for edits. Before that boundary, check the current question's contract, required files or explanation, and visible test results.

For our practice task, your final review should answer four concrete questions: Does the function return the requested structure? Does it keep the newer event even when input is out of order? Does it implement the tie rule? Does malformed input leave valid winners intact?

If time is limited, prioritize a correct baseline and a test that challenges its central assumption. Avoid replacing a tested scan with an unverified optimization immediately before submission. If you do change the code, rerun the case that originally established correctness.

Recommendation: Treat question submission and interview completion as distinct checkpoints. Complete any requested explanation or final information step, then inspect the platform's completion state. A passing run, an auto-save event, and a submitted question each establish less than a completed interview.

Recover from a problem without guessing about the timer

Official fact: The FAQ describes auto-saving progress and reopening the original invitation link to resume after a browser interruption. It also directs candidates to the hiring company for results. That does not promise that a timer resets, every last keystroke survives, or an employer extends a deadline.

Recommendation: If the editor fails, record the symptom and time, preserve the visible error when appropriate, and use the official support route. Distinguish a hardware or loading problem from uncertainty about the task. Ask the recruiter about deadlines, permitted assistance, or ambiguous requirements; ask platform support about technical access.

A useful problem report describes what stopped working, which step you had reached, and whether a question had already been submitted. Avoid repeatedly restarting a timed task without understanding the current state.