Turn interviews into hands-on sessions

A technical interview is the step where the company watches the person solve a real technology problem instead of just talking about their experience. A good technical interview assesses four things: reasoning (how the person breaks the problem down), code (quality, organization and technical choices), communication (how they explain decisions and trade-offs) and architectural depth when the role requires it. On Coodesh, these four areas can be covered in a single assessment, combining hands-on coding questions, a whiteboard for system design and answers by video, audio or a conversation with an AI interviewer.

In live coding, an engineer from the company follows the candidate in real time while they code. In an async interview, the candidate solves the problem in their own time within a set window, and the evaluation happens afterwards based on what was submitted and recorded. Live coding gives more signal on how the person thinks under pressure and costs one senior engineer hour per candidate. Async scales to dozens of candidates and frees the technical team to interview only finalists. Coodesh is designed so the async step does the heavy lifting of technical screening, and the human conversation happens with people who have already shown they can do the work.

By making it a criterion instead of pretending AI doesn't exist. Developers use coding assistants on the job, so the useful question is no longer whether they can code without help but whether they know how to prompt, review, fix and explain what the AI delivered. On Coodesh, the company decides per assessment whether AI use is allowed and writes the criteria accordingly: quality of the final solution, ability to justify decisions in the video explanation or the conversation with the interviewer, and spotting errors in the code.

With evaluation criteria (a scorecard) defined before the interview and automatic review applied to them. For each open question, the company writes weighted criteria; the platform scores each criterion, generates a summary and points to evidence in the candidate's submission. The human reviewer arrives with the reading done: they confirm, adjust the score where they disagree and add comments. What used to take 40 minutes of review per candidate takes a few minutes, and two different reviewers use the same yardstick. The decision stays with the person.

Integrity signals notify interviewers when the system detects meaningful clusters of suspicious candidate activity, such as repeated copy-paste, frequent window resizing and tab switching. Coodesh alerts interviewers to tab switches, copying and pasting code, multiple monitors and help from any third-party tools, even those that claim to be invisible. Integrity monitoring runs passively in the background, so interviewers stay focused on the conversation instead of policing behavior.

With photo ID and a continuous record. Coodesh lets you require a photo of the candidate at the start of each assessment and periodic webcam snapshots throughout, with automatic analysis that flags an absent face, more than one person or a change of person. Because the technical test and the technical interview live on the same platform, the reviewer can compare the record of each step for the same candidate without another tool.

The technical screening test is objective and scales: quizzes, coding challenges graded by unit tests and projects, applied to every candidate to filter who has the required foundation. The technical interview is open-ended and deep: free-form coding without automated tests, whiteboard and real-time explanation, applied to those who passed the screen to observe reasoning, communication and decisions. Both live in the same assessment or in chained assessments, with the same integrity features, and the candidate doesn't have to learn a new tool between steps.

Most hiring processes still ask candidates to implement an algorithm, testing memorization of algorithms that AI can generate in seconds. On day one, a senior engineer opens VS Code with Copilot or Claude, picks up a vague Jira ticket, uses AI to create the basic structure, critically reviews what the AI produced, finds where it's wrong or incomplete and writes the most complex parts on their own. Coodesh was built to test exactly these skills: AI fluency, code quality, critical thinking and real-world problem solving, in an environment that mirrors how developers actually work.

Coodesh coding questions cover the main languages on the market, such as Python, JavaScript, TypeScript, Java, C#, Go, Kotlin, Swift, PHP and C/C++, among others, and the library maps more than 30 careers, from frontend and backend to data and DevOps. The assessment can be set up in Portuguese, English or Spanish, so you can use the same process for roles in Brazil, Latin America and beyond. If you need a stack or scenario that isn't in the library, you can create your own question or ask the Coodesh team.