Companies Are Using AI to Interview You — And Banning AI While They Do It

There is a quiet contradiction running through every technical interview being scheduled right now. On one side, companies are using AI to screen candidates: McKinsey is rolling out a ten-minute AI simulation of a client conversation for some entry-level candidates, asking applicants to demonstrate judgment and communication without a human in the room. On the other side, those same companies — and many beyond consulting — are adding proctoring software, camera-on requirements, and explicit bans on AI assistance during their coding assessments.

You are expected to be AI-fluent at work. You are expected to be AI-free during the test. That tension is not an accident or an oversight. It is a signal about what employers actually want, and understanding it is one of the clearest advantages a job seeker can claim heading into the second half of 2025.

Why the Cheating Crackdown Arrived So Fast

The catalyst most people in hiring circles point to is a browser extension originally called Interview Coder, later rebranded as Cluely, that was built to feed candidates AI-generated answers during live coding screens and video calls. The backlash inside hiring teams was immediate. Companies including Visa Inc. began adding explicit prohibitions on AI tools and third-party assistance to their assessment policies. Proctoring steps that did not exist in most pipelines eighteen months ago are now standard at a growing number of firms.

The logic is straightforward: when an AI coding assistant can clear a standard algorithm screen with minimal human input, the screen stops measuring what it was designed to measure. Companies responded by hardening their formats — more camera-on sessions, more collaborative debugging exercises, and a shift toward system design discussions that are genuinely harder to outsource to a tool in real time.

The practical consequence is blunt: you need to be able to perform cleanly under proctored, no-assist conditions. If your practice routine relies on having Copilot or Claude open in a side window, you are training for a test that now has stricter rules than the one you will actually face.

The Other Side: AI Is Also the Interviewer

While the proctoring pressure increases on candidates, some employers are deploying AI on their own side of the table. McKinsey's AI-simulated client interview is an early and visible example — a ten-minute conversation with a system acting as a business stakeholder, designed to evaluate how candidates handle ambiguity, ask clarifying questions, and communicate under low information conditions. This is a structurally different kind of assessment from a LeetCode screen. It rewards fluency with AI, not circumvention of it.

This two-sided reality is reshaping what a strong candidate looks like. Employers want engineers who can think and ship without a crutch, and who can also work fluidly alongside AI tools in their actual day-to-day practice. Those are complementary skills — but you have to train for each deliberately, because the interview contexts that surface them are increasingly separate.

What the Demand Data Is Telling You

The broader market is not making this easier. July 2025 brought 62,075 job cut announcements across the United States — a 29% jump from June — with technology, AI-driven automation, and tariffs cited as the leading causes, according to Challenger, Gray & Christmas. Microsoft alone cut roughly 9,000 positions in July, its second significant round of reductions this year. The job platforms most candidates rely on are not insulated either: Recruit Holdings eliminated 1,300 roles across Indeed and Glassdoor as the parent organization accelerates its pivot toward AI-powered job matching.

In a tighter market, specificity in hiring has increased sharply:

Role category	Trend vs. February 2020 baseline
Machine learning engineers	+59% above baseline
Overall US tech job listings	−36% below baseline

Source: Indeed Hiring Lab data via Pin.com's tech job market report

The gap between those two numbers is not a coincidence — it reflects a deliberate reallocation of engineering headcount. Companies that are still growing their teams are growing them in a specific direction: toward people who can build, evaluate, and integrate ML systems, not just maintain existing software. Even one meaningful ML project on your profile — fine-tuning an open model, calling an LLM API to build a tool, running inference with PyTorch — changes where your resume lands in a competitive stack.

For engineers on H-1B visas, the picture carries additional urgency. Employers are growing more selective about sponsorship, with demonstrated AI tool proficiency becoming a threshold criterion in many evaluations. Separately, starting September 2, 2025, the State Department has rescinded the broad policy that allowed consular officers to waive in-person interviews — H-1B, L-1, and O-1 holders must now show up in person at a consulate as the default, not the exception. If a renewal or stamping is on your horizon, book your consulate appointment now before wait times compound with the new backlog.

How to Prepare

1. Train without the assist, on a schedule. Set a dedicated weekly block to solve algorithm and data structure problems with no AI tool open. The goal is not to avoid AI — it is to know what you can do without it, so proctored conditions do not catch you off guard.

2. Build your system design fluency now. Leaner teams mean the engineers who do get hired are expected to operate at a higher architectural level. Expect more system design questions and fewer pure algorithm screens in loops at companies that have recently cut headcount. freeCodeCamp's free Learn Software System Design course covers databases, scaling, APIs, and security in about two hours — a practical starting point.

3. Add one real ML or AI integration project to your portfolio. Put it on GitHub with a clear README, and describe it on your resume with a measurable outcome — lines of code is not a metric; latency reduced, accuracy improved, or time saved is.

4. Name your AI tools explicitly on your resume. If you use GitHub Copilot, Cursor, or similar tools in your actual work, list them in your skills section and describe the productivity impact in project entries. This is especially important for visa-sponsored candidates competing in a more selective sponsorship market.

5. Practice talking through your reasoning out loud. Both proctored coding screens and AI-simulation interviews reward candidates who can verbalize their thinking in real time. Record yourself solving a problem or walking through a design trade-off, and watch the replay — most engineers are surprised by how differently they reason aloud versus on paper.

6. Diversify your sourcing beyond Indeed and Glassdoor. With both platforms under structural pressure from their parent company's own AI pivot, do not treat them as your primary pipeline. Add LinkedIn, company career pages, Dice, and Wellfound to your active rotation. Ryzlink's current engineering openings — across contract and direct roles — are listed and updated at ryzlink.com/jobs, and Ryzlink typically matches qualified engineers to client teams within 48 to 72 hours.

Free Courses to Get You Ready

Design and Analysis of Algorithms (6.046J) — MIT OpenCourseWare This course works through the core algorithm techniques — divide-and-conquer, dynamic programming, greedy methods, and cryptographic foundations — that show up repeatedly in proctored coding screens. Cost: Free. Lecture videos, notes, recitations, and problem sets are all accessible without creating an account.

Data Structures and Algorithms (Full Course from a Google Engineer) — freeCodeCamp / YouTube A complete walkthrough of data structures and algorithms taught by a Google engineer. Cost: Free. Open the link and press play — no account needed.

Learn Software System Design — freeCodeCamp Two hours of system design fundamentals spanning database design, scaling strategies, load balancing, RESTful and GraphQL APIs, and security considerations — directly relevant to the architecture-level questions now appearing in more interviews. Cost: Free. Open the page, click the YouTube link — no account needed.

CS50's Introduction to Computer Science — Harvard / edX For career changers or new graduates who want to establish genuine foundational fluency across C, Python, SQL, and web development. Cost: Free to audit; optional paid verified certificate. Visit cs50.harvard.edu/x to get started; set up an edX account if you want to submit assignments and receive feedback, and use the edX enrollment option if a verified certificate is your goal.