How to Get a Job as an AI Product Designer

Here's something that became obvious after reviewing 300+ applications for Google DeepMind: the design industry has quietly split into two tracks, and most designers are on the wrong one.

Track one is the traditional path. Great portfolio, clean case studies, solid experience at recognized companies. Five years ago, this got you into almost any design role. Today, at DeepMind, it gets you rejected at the profile review stage.

Track two is what we're calling the AI native multi-tool designer. And it's becoming the new baseline, not just at DeepMind, but across the companies building the most interesting products right now.

The numbers are brutal

We've presented eight to ten candidates for these roles. Three made it to a first round interview. The rest, people with extraordinary resumes and genuinely great work, were rejected before they even got a phone call.

The reason was consistent: nothing moving. No vibe coding. Static portfolios, no matter how polished, are not clearing the bar.

And this is Google DeepMind. The Gemini team. The team that most designers in the world would consider the pinnacle. When they set their hiring bar, it's a directional indicator for the rest of the industry.

What "AI native" actually means (from the hiring managers)

We got direct feedback from DeepMind's hiring managers on exactly what they evaluate. It breaks down to three things:

Tough problems, not polished deliverables. They want staff-level problems. Complex, ambiguous, high-stakes. They want to see how you think through the mess, not just the clean output at the end. Preferably showcasing AI in the work, but not required. The emphasis is on the quality of the problem and your reasoning, not the presentation veneer.

High-craft visual and interaction design. The aesthetic bar hasn't dropped. It's risen. But it needs to be paired with sharp interaction thinking and, critically, things that move. Even a beautiful side project counts. Something animated, 3D, interactive, boundary-pushing. Craft is table stakes, and you need it alongside everything else.

Concise storytelling. If you can't tell your story without taking forever, you're out. These teams move fast. They need people who can communicate complex ideas clearly, under pressure, without meandering.

Demos over decks: the practical reality

There's a "demos over decks" mentality on this team, and it's not just philosophical. There's a practical reason too.

When you're presenting via video call, those looping video presentations in your portfolio deck tax your GPU and theirs. The screen stutters. The audio desyncs. The experience for the interviewer is actually terrible, even when the work looks gorgeous in theory.

But when you walk through a live prototype hosted on Vercel or Netlify, it's lightweight. It's your voice plus the visual. They can feel the craft. And if they want to click around themselves, they can.

The takeaway: for your async portfolio (the website you send ahead), embedded videos are fine. For the live presentation, a working demo you can walk through while explaining your decisions is significantly more powerful.

The workflow chain is the flex

Here's what separates track two designers from track one: it's not just the output. It's showing how you got there.

A candidate we recently worked with did this brilliantly. They had an interaction design project, an audio waveform component. Here's their workflow:

1. Threw the concept into an image generation tool for initial visual exploration. Results were mediocre, but the broad strokes were right.
2. Took that direction into a different AI design tool for more refined stylistic variations.
3. Moved the code into Cursor: "Help me build this, with this visual direction."
4. Needed it in Framer, so had Cursor output the code as a framework component for direct paste.

At no point did they open Figma.

The designer's eye showed up during the Cursor phase. Adjusting easing. Tweaking proportions. "The audio shouldn't be so reactive." "There needs to be some easing." "Too thick. Too wide." That's where design taste meets technical capability, and that's the skill these teams are hiring for.

The move for your portfolio: devote a slide or a section to your workflow before showing the work. Talk about your tool chain. Talk about how it's changing. Talk about how you decide which tool to use for what. Even just saying "my process changes every week and I'm constantly adapting" is a strong signal because it's true, and it shows you're in the arena.

What to build if you need something new

If you're going to build from scratch, think about the problem space these teams operate in.

The Gemini Connected Apps team is thinking about how first-party Google apps (Drive, Sheets, Docs) and third-party tools (Asana, GitHub, Jira) integrate into the Gemini experience. They're thinking about how rich, interactive components surface inside an AI conversation. They're thinking about multimodal experiences across web, iOS, and Android.

You don't need to design for Gemini specifically. But building something that demonstrates you can think about connected ecosystems, multimodal interfaces, or AI-native interaction patterns at that level of complexity will land much better than another e-commerce redesign.

And the time investment might be smaller than you think. We've seen designers rebuild old Figma projects as working Cursor prototypes in a single evening. Three hours. The gap between "static portfolio piece" and "interactive portfolio piece" has collapsed. The tools are there. The question is whether you're using them.

Why this matters beyond DeepMind

This isn't just a Google thing. We're seeing the same shift across our client base. The designers getting hired at the highest levels right now are the ones who can move between design tools, code environments, and AI workflows without friction. They prototype fast. They think in systems. They ship things you can touch.

Here's what we expect over the next 12 to 18 months:

• More companies requiring prototyping as part of the design challenge, not just a presentation
• Portfolio case studies that include working demos outperforming those that don't
• "Design engineer" and "AI native designer" becoming standard role titles, not niche ones
• The compensation premium for designers who can prototype continuing to widen

The old model of "I design it, engineering builds it" is compressing rapidly, especially on AI product teams where the design space is so new that you need to explore by building.

How to get on track two

Start with one project. Take a design you've already done and rebuild the core interaction as a working prototype. Use whatever tool gets you there fastest. Cursor, Bolt, Replit, v0, doesn't matter. The point isn't code quality. It's proof that you can close the gap between idea and artifact.

Then put it in your portfolio. Make it the first thing people see. Not a video of it, not a screenshot. The actual thing. Host it. Let them play with it.

Then talk about how you made it. That's the second flex. The workflow story is almost as valuable as the output itself.

That's what DeepMind is asking for. That's what the best teams everywhere are starting to ask for. And that's the new baseline.

Academy UX is the exclusive recruiting partner for Google DeepMind's design team. If you're an AI native designer ready for the hardest design problems in the world, apply here.

B150 is a professional community for GenAI builders. Join us at b150.ai.