See candidates write and run code before your engineers interview them
Candidates solve coding problems in a browser editor, in the language they use at work. Hidden test cases check each answer, and every run is saved. Your engineers only interview people who solved the problems.
rate_limiter.py
Wei Chen · Run 3
1def allow(q, now, limit, window):2 while q and q[0] <= now - window:3 q.popleft()4 if len(q) >= limit:5 return False6 q.append(now)7 return True
- Sample 1
- Sample 2
- Sample 3
Free · Email code to start · Uses your mic · Up to 3 minutes
Trusted by HR teams at 300+ companies
Why engineers dread first-round interviews
A senior engineer can lose a day a week to first-round calls. Many of those candidates cannot write a working loop under time pressure. A take-home task takes days to come back, and someone else may have written it. A timed test in the browser shows within the hour who can write code that runs.
What your engineers see before the first call
The code, every run, the hidden test results, the integrity log and an AI interview on design, all in one report.
Code editor
Candidates write and run code in the language they use at work
The test opens in a browser editor. Candidates pick a language you allow, such as Python, Java, Go or TypeScript, and run their code against sample inputs. Pick problems from a library of about 1,000, from arrays to SQL and Bash, or write your own.
top_customers.sql
Sofia Rossi · SQL · Run 2
1SELECT c.name, SUM(o.total) AS spent2FROM customers c3JOIN orders o ON o.customer_id = c.id4WHERE o.created_at >= '2026-01-01'5GROUP BY c.name6ORDER BY spent DESC7LIMIT 5;
- Sample 1
- Sample 2
Hidden tests
Hidden test cases check inputs the candidate never sees
Sample inputs only show the basic case. On submit, Goodfit runs hidden tests for edge cases like an empty list, negative numbers or a very large input. A problem counts as solved only when every hidden test passes.
Merge intervals
Daniel Okafor · Java
- Overlapping pairsPassed
- Empty inputPassed
- Negative start timesFailed
- 100,000 intervalsTimed out
- Already mergedPassed
The candidate sees pass or fail, never the hidden inputs
Run history
Every run is saved, so you can see how they fixed a bug
Each run and submit is kept with its code, result, runtime and memory. Compare any run with the one before it, the first try or the final answer. You can tell whether the candidate worked through a bug or started over.
Runs · Rate limiter
Wei Chen · Python · problem 2 of 4
Run 1
1 of 310:04 · Counted requests per second instead of per window
Run 2
2 of 310:15 · Fixed the window reset
Run 3
3 of 310:23
Submit
8 of 8 hidden10:31 · Solved
Integrity
Watch the webcam and jump to each flag
Goodfit records the webcam and a replay of the test page. Flags mark tab switches, blocked pastes and developer tools being opened, each with the time it happened. If an AI browser add-on like HARPA or Monica was open, the report names it.
Recording- Developer tools opened08:10
- Paste blocked: 1,180 characters18:05
- Switched tabs for 2 min 10 s31:40
- HARPA add-on in use44:02
Pass mark
Set how many problems a candidate must solve to move on
The score is the number of problems solved out of the number given. You set the pass mark for each test, like 3 of 4. Candidates below it are marked poor fit, and your team can still open any report.
Results · Backend coding test
Pass mark: 3 of 4 solved
- WC4 of 4
Wei Chen
Python · 48 min
- SR3 of 4
Sofia Rossi
Java · 55 min
- LM2 of 4
Lucas Meyer
Go · 60 min
- DO2 of 4
Daniel Okafor
Java · flags to review
AI interview
An AI interview asks about system design and past projects
After the coding test, the AI asks how the candidate would design a service, what they would trade off and what broke in a past project. It follows up when an answer stays vague. Your engineers keep the final round for the hardest questions.
AI interview · Backend Engineer
Sofia Rossi · Browser
Live transcript
How would you design a service that sends 2 million push notifications in an hour?
I would put the jobs on a queue and run several workers.
Follow-up
What happens if a worker dies halfway through a batch?
Each job is marked done only after the send, so it goes back on the queue and retries.
Free to start
Try Goodfit on your next open role
Create a free account, add a job and send the AI interview to real applicants. You get a scorecard for each one with the recording behind every score.
- 20 free assessments
- No card needed
- Set up in minutes
DBDaniel Brooks
Your first applicant · scored minutes after the AI interview
8.0 /10
- Explains clearly8/10
“First I’d check the order, then call the customer back.” ▶ 02:14
- Handles pressure8/10
“I stayed calm and told them exactly what I could do today.” ▶ 05:40
Customer story · Logile
Logile hires backend, frontend, mobile and QA engineers with a three-person TA team. A proctored Goodfit interview replaced one full round of live interviews and cut shortlisting and screening time by 25 to 30%.
“Every resume that comes through today is essentially AI-generated, so on paper it looks perfect and matches the job description word for word. Last week alone we interviewed a dozen candidates who all looked exactly right, and every single one fell short once we actually spoke to them.”
25–30%
Less time on shortlisting and screening
1,785
Applications processed since launch
FAQ
Questions about coding tests on Goodfit
Which languages can candidates use?
Do partial answers earn points?
Which SQL do candidates write?
Can we set a time limit?
Let your engineers interview only people who can code
Pick four problems from the library or write your own. Your engineers only interview candidates who passed the hidden tests.
- 20 free assessments, no card needed
- Then $5 per assessment
- No seats or annual contract