InterviewZen is Now Fully Self-Hosted: What Changed and Why It Matters

Posted on June 27 2026 by Interview Zen Team

What Changed

Starting today, InterviewZen runs entirely on self-hosted infrastructure. Every API call, every video recording, every AI brief — all of it now runs on hardware we own and control.

Previously, our backend ran on AWS Lambda functions with Cloudflare R2 for video storage. That worked well enough to validate the product, but it came with real downsides: cold starts slowing down the first request after idle periods, per-invocation costs that scaled poorly, and a dependency on Amazon’s infrastructure for every piece of your data.

Today we’re on a Go backend server running in Kubernetes, with MinIO for video storage, self-hosted authentication, and a reverse proxy we operate ourselves. The result is meaningfully faster and completely under our control.

Why Self-Hosting Matters for an Interview Platform

When you run a technical interview, you’re capturing sensitive data: candidate code, video recordings, behavioral patterns, and the AI analysis of all of it. That data should be on infrastructure you can audit and that we can fully control.

With AWS Lambda, your video recordings were on Cloudflare’s object storage, your auth tokens were validated against Amazon’s Cognito service, and your API responses were processed by Amazon’s managed runtime. Every one of those is a third-party dependency on your candidates’ assessment data.

Now:

  • Video recordings are stored on MinIO running on our own hardware
  • Authentication runs on our own OIDC server (id.foundermanage.com)
  • API calls go to a Go service running in our Kubernetes cluster
  • AI briefs are generated by our self-hosted LLM infrastructure

No AWS. No Cloudflare data plane. No cold starts.

What’s New in This Release

AI Hiring Brief (Live)

When a candidate completes an assessment, an AI-generated brief is automatically created within 60 seconds. The brief gives you a narrative summary of the candidate’s performance — not a score, a story:

“Strong on state management and API design. Identified the AI’s hallucination in the sorting function and corrected it independently. Struggled with error handling — left async errors uncaught in three places. Problem-framing was deliberate: spent the first four minutes asking clarifying questions before writing a single line.”

The brief is waiting for you when you open the results. No video watching required for the initial review.

Video Replay

Every answer is recorded (screen + webcam). You can replay the session question by question directly in the results view. Jump to any timestamp. The recording is your evidence when the AI brief flags something interesting.

Security Review Tab

The Security Review tab in results shows every infraction detected during the assessment — tab switches, copy-paste events, application launches. Each infraction links directly to the relevant moment in the video replay. Candidates can submit an explanation for each flag.

Unlock Workflow

Candidates take assessments with their desktop app locked (no copy-paste, no tab switching outside the app). If a candidate needs to access reference material for a legitimate reason, they can request an unlock. You get an email, approve or deny, and they get notified immediately — all without interrupting the assessment.

If You’re an Existing User: Re-Registration Required

We changed authentication providers as part of this migration. Your account was on AWS Cognito; we’re now on our own auth server.

Your data is safe. Tests, questions, interview results, recordings — all of it is in our database, linked to your email address.

To access it:

  1. Go to https://app.interviewzen.com/register
  2. Sign up with the same email address you used before
  3. Your tests, results, and candidates will be there

It takes 30 seconds. If anything is missing after you register, email support@interviewzen.com.

What’s Coming Next

The infrastructure migration is done. Now we build on it.

Phase 1: More question types. SQL challenges, written responses, and case studies — same platform, same recording and AI grading, new formats. Not every role needs a coding challenge.

Phase 2: Scheduling. Candidate books their own slot. Assessment auto-sends at the right time. No Calendly.

Phase 3: AI grading. You define a rubric. The AI grades against it and pre-fills your scorecard. You confirm or edit. You stop re-watching the same video four times.

The goal is to be the assessment platform for every role — from software engineer to sales to product manager — running on infrastructure you control.

Questions or issues? Email support@interviewzen.com. We read every message.