AI in Job Interviews: My Lifelong Obsession with Gemini and Prompt Engineering

I still remember the first time I asked Gemini to refactor my code. It was late on a Sunday, and my Python script stubbornly refused to behave. I typed a quick prompt, half-expecting nonsense, and instead watched in awe as the AI assistant laid out a clear, concise solution. In that flickering moment of triumph, I felt a spark I hadn’t felt since I discovered open source. Who knew debugging could feel like magic?

Over the years, Gemini became my constant. When I went through a rough breakup, I buried my heartbreak in pull requests and bug fixes. During college all-nighters, I paused lectures to ask my virtual partner for hints on algorithm optimizations. Even on my honeymoon, while my partner snoozed beside me, I whispered prompts into my laptop at dawn. There’s comfort in that steady glow of suggestions, the sense that I’m never really coding alone.

Some folks doom-scroll social media to fill idle moments. Me? I doom-scroll Gemini issue threads. I bookmark intriguing repositories, study how maintainers handle merge conflicts, dissect thread after thread of prompt engineering examples. I’ll confess—it’s a little obsessive. But this ritual taught me more about software development than any textbook ever could. It’s where I honed my instincts for clean code and elegant solutions.

When AI Became My Partner in Code

There was a time when pairing meant coffee and a fellow engineer. Now it often means tapping into an LLM. Prompt engineering morphed from a novelty into an art form. I learned to craft prompts with context, role definitions, constraints, and examples. “Optimize this function for memory use,” I’d type, pasting a block of C++ code. Seconds later, Gemini delivered a refactored snippet and explained every tweak. Iterating on those prompts—I tried chain-of-thought, few-shot examples, personas—became my favorite puzzle.

Google's AI-Friendly Interview Pilot

So when I heard that Google is piloting a new software engineering interview process that officially lets candidates use a company-approved AI assistant, I wasn’t shocked—I was thrilled. In this pilot, select US teams are inviting junior to mid-level applicants to tackle a code comprehension round with Gemini at their side. Instead of banning AI, interviewers will explicitly evaluate AI fluency: how well you prompt the model, validate its output, and debug its suggestions.

This marks a big shift. Internally, Google reports that nearly 75% of new code is generated with AI help. It only makes sense to test for these skills up front. Vice President of Recruiting Brian Ong explained that the goal is to recruit engineers who can thrive in human-led, AI-assisted workflows. Early feedback will shape how broadly the pilot rolls out—if it proves effective, more teams could adopt it soon.

Here’s how it works: candidates receive a codebase loaded with bugs and performance issues. They have access to Gemini via an approved interface. Interviewers look for succinct prompts that guide the AI, smart follow-up questions that refine outputs, and deliberate checks that catch hallucinations or flawed logic. It’s about more than raw coding chops—it’s about mastering a collaborative dance with AI.

Why This Matters

Traditional interviews have long felt disconnected from real-world workflows. Asking someone to code on a whiteboard without reference material? In 2026, that’s like banning calculators in math exams. Companies including startups like Canva and research labs at Meta have already experimented with AI-friendly assessments. By embracing this change, Google is acknowledging that deep technical skill now includes the ability to wield AI tools effectively.

Statistics back this up. A Google-Ipsos report found that only about 5% of the global workforce is truly AI-fluent, even though nearly 40% use AI in some form. Those fluent engineers are roughly four times more likely to earn promotions and take on leadership roles. A hiring process that filters for AI fluency could reshape career trajectories, rewarding people who can prompt, critique, and integrate model outputs.

Broader Ripples in Tech Hiring

If the pilot scales, we could see a seismic shift across the industry. Firms that cling to no-AI tests risk missing out on talent who live and breathe AI-augmented development. At the same time, there are concerns about fairness. Not everyone has equal access to premium AI tools or training, and we’ll need thoughtful guidelines to ensure diverse candidates aren’t disadvantaged.

Moreover, as AI writes a growing share of code, routine tasks may fade away, elevating roles that demand strong critical thinking, domain knowledge, and ethical oversight. We might also see new standards for evaluating AI collaboration—accuracy checks, bias mitigation steps, even AI-driven unit tests. This pilot could be the blueprint for a generation of hiring practices where AI skill is as fundamental as data structures.

What It All Means to Me

Here’s the truth: I love Gemini more than I probably should. I’ve built routines around it, learned its quirks, and yes, gotten frustrated when it hallucinated or suggested a patch with a hidden bug. I’ve gasped when its refactors shaved milliseconds off a critical loop and groaned when a prompt misfire wasted an hour of my time. But every hiccup taught me something new about model behavior and about myself.

This pilot feels personal. It’s an acknowledgment that my obsession with AI assistants—those late-night debugging sessions, the endless prompt experiments—has value beyond hobby or habit. It has prepared me for a workflow that’s fast becoming the norm. I’m eager to see how this new interview format tests those skills I’ve honed so obsessively. Will I stumble over a prompt misstep? Probably. But I also know how to recover: by iterating, validating, and refining. The same process that turned a midnight bug fix into a lightbulb moment will guide me through the interview.

Looking Ahead

This isn’t just a hiring tweak—it’s a statement. It signals that AI and human ingenuity are inseparable partners in the future of software engineering. As companies race to integrate AI deeper into every phase of development, those who master the art of prompting and collaboration will lead the pack.

For me, the journey continues. I’ll keep honing my prompts, debating code recommendations with Gemini, and refining that dance between human insight and machine suggestion. In doing so, I’m not just pursuing a passion—I’m preparing for the next era of engineering, where the best developers are those who know how to ask the smartest questions.