Vetting
How to see real developer test scores before you interview
Ask for proctored test scores, not self-reported skills, and read them before you book the interview. A trustworthy score comes from a timed test that a real person watches, on work that looks like the job, given the same way to every candidate. On Arca.ph, profiles are free to view, any test scores already on file sit on the profile, and your own AI assistant (Grok Bot, Claude or ChatGPT) can sort candidates by those results before you talk to anyone.
This page explains what makes a score worth trusting, when to run your own assessment tool instead, and the steps to see scores before the first interview.
What makes a developer test score trustworthy?
A score is only useful if you can trust how it was produced. Check for four things:
- It was proctored. A real person watched, or the session was recorded with screen and camera, so the work is the candidate's own. Without that, a modern AI assistant can quietly do the test for them. TestDome's own guide says no single feature makes a test AI-proof, and that a hidden overlay assistant or a second device needs a hard lockdown and webcam proctoring to catch (TestDome).
- It looks like the job. The Office of Personnel Management (OPM), which sets hiring guidance for federal agencies, describes work sample tests as tasks that "mirror the tasks employees perform on the job," and notes that performance on them "relates highly to performance on the job" (OPM, work samples).
- It was timed and the rules were clear. You should know whether AI tools were allowed. A score from a test with no AI tells you something different from a score on a build project where AI was allowed. Both are useful; mixing them up is not.
- Every candidate took the same test. Comparing scores only works when the test and the scoring are the same for everyone. OPM makes the same point about interviews: structured interviews ask every candidate the same questions and score answers against the same standard (OPM, structured interviews).
A resume line that says "expert in Python" passes none of these checks. A badge from an unproctored quiz passes some of them. A proctored, timed work sample passes all four.
Should I use a developer assessment tool or read scores that already exist?
Both are valid, and they solve different problems.
Run your own assessment tool (HackerRank, Codility, TestDome and similar) when:
- You already have a pipeline of applicants and need to screen them in volume.
- You want a test written around your exact codebase or stack.
- You have someone on your team who can set up the test, choose the proctoring options and review the results.
Read scores that already exist when:
- You haven't found candidates yet, so there is nothing to send a test to.
- You'd rather compare people on the same proctored tests than write your own.
- You want your AI assistant to sort a large set of profiles by evidence, not by keywords.
The question "what are the best developer assessment tools?" usually hides a more useful one: "how do I see real evidence before I spend an hour in an interview?" An assessment tool is one answer. A marketplace where proctored results sit on the profile is another.
Which proctored tests tell you the most?
Pick the test that matches the risk you're worried about:
- English and thinking-skills tests. Useful when the role involves a lot of written communication with your team.
- A coding test with no AI allowed. Shows whether the developer can write and reason about code without help.
- A 2-hour build project with AI allowed. Shows how they work the way most teams now work: using AI tools, then checking and fixing what the tools produce.
- The Fire Drill. A live pair-programming session on a real-world problem. The developer talks through their choices while they build, so you see how they think, not just the final answer.
- Recorded reference checks. A former manager or colleague answers questions on a recording you can review.
For more on the no-AI test, see How to tell if a developer can code without AI. The full list is on Proctored tests.
How do I see test scores before the interview on Arca.ph?
- Get started. Create an account at app.arca.ph/sign-up with your work email. Profiles are free to view.
- Write a short brief. List the skills, the pay range (per month), the hours you need to overlap, and which test results matter for this role.
- Ask your AI assistant to search and sort. Grok Bot, Claude or ChatGPT searches the profiles and sorts them by the evidence you care about, such as a no-AI coding score or a Fire Drill result. Your assistant builds the shortlist. Arca.ph doesn't shortlist for you.
- Ask for an introduction, then ask for proof. Tell Arca.ph who you want to meet. If the developer says yes, that's an introduction. If a profile doesn't have the result you need yet, ask for the proctored tests or recorded reference checks you want. Arca.ph returns the scores to you.
- Interview only the people whose scores hold up. Use the interview to check fit and to dig into the test, for example "walk me through why you chose that approach in the Fire Drill."
- Hire and pay once. If you hire someone Arca.ph introduced, within 1 year of the introduction, you pay one fee of 15% of first-year pay, once. Every hire includes a 3-month replacement.
Tests are paid and optional. Looking at profiles stays free.
What does a hire cost?
On the marketplace, one success fee of 15% of first-year pay, charged once, only if you hire someone Arca.ph introduced within 1 year of the introduction. Pay is quoted per month, up to about $3,000 a month. For example, a developer paid $3,000 a month is $36,000 for the first year, so the fee is $5,400, once. See Pricing.
If you'd rather not run the search yourself, Done-for-you is 35% of first-year pay, charged once on hire.
Get started or Sign in.
FAQ
How can I see real test scores for a developer before I interview them?
Ask for proctored, timed work-sample results instead of self-reported skills. On Arca.ph, profiles are free to view and test results on file sit on the profile, so your AI assistant can sort candidates by them before you book an interview. After an introduction, you can ask for more proctored tests and Arca.ph returns the scores.
What are the best developer assessment tools?
If you already have applicants and want to test them yourself, tools like HackerRank, Codility and TestDome let you build and send coding tests. If you haven't found candidates yet, a marketplace where proctored results are already on the profile saves you writing and running the test.
Why does proctoring matter for a coding test?
Without someone watching, an AI assistant can do much of the test for the candidate. A proctored test records the screen and the person, so the score reflects the candidate's own work, with or without AI depending on the test's rules.
Is a no-AI coding test or an AI-allowed project better?
They answer different questions. A no-AI coding test shows fundamentals. A 2-hour build project with AI allowed shows how someone works with modern tools and checks their output. Many teams look at both.
Does Arca.ph pick the candidates for me?
No. Your own AI assistant searches and sorts the profiles, and you decide who to meet and who to hire. If you want Arca.ph to run the search, that is Done-for-you, at 35% of first-year pay, charged once on hire.
How much does it cost to see profiles?
Nothing. Profiles are free to view. Proctored tests are paid and optional. If you hire someone Arca.ph introduced within 1 year of the introduction, you pay 15% of first-year pay, once.
Sources
- Office of Personnel Management (OPM), "Work Samples and Simulations": https://www.opm.gov/policy-data-oversight/assessment-and-selection/other-assessment-methods/work-samples-and-simulations/
- Office of Personnel Management (OPM), "Structured Interviews": https://www.opm.gov/policy-data-oversight/assessment-and-selection/structured-interviews/
- TestDome, "How to Stop AI Cheating in Coding Assessments": https://www.testdome.com/guide/ai-cheating-coding-assessment
- Arca.ph, Pricing: https://arca.ph/pricing