A Diagnostic of junochen.com
Your published thesis, across every case on your site, is that humans need visibility, verification, and control before they trust a consequential system with authority. Alibaba. Red Cross. Thermo Fisher. An entire essay called "Trust Is the New Interface."
A hiring manager evaluating you for a Head of Design role is performing exactly the assessment you design for. Can I trust this agent with authority? Do the signals match what's labeled? Does the system behave the way it claims to?
That is the lens for everything below. Does this trust surface pass its own test? Where it does, I'll say so. Where it doesn't, I'll tell you what to fix and what to say if it comes up before you fix it.
The level signal is already there
Start here, because the anxiety behind this question may be solving a problem that no longer exists on the live site.
The current homepage reads "Design leader for platforms where trust is the product." The page title says "Juno Chen — Design Leadership for High-Stakes Systems." The meta description says "Head of Design." The eyebrow line says "Available · Design Leadership · AI-native platforms · San Francisco Bay Area." None of this is IC language. A recruiter landing on this page gets Director+ signaling within the first scan.
The case teasers reinforce it. Alibaba: "Head of Design & Research, North America," 32 research interviews, executive buy-in. Red Cross: "Product Design Director / GM · 0→1 Build." Thermo Fisher, Equinox+, and Allē all carry "Product Design Director." Every teaser leads with mandate scope. The verbs are "led," "shipped," "built," "secured." Leadership verbs attached to leadership titles at recognizable companies.
The TinyFish title question is handled in the hero copy itself. The framing is honest and forward-facing: understanding product decisions makes you a sharper design leader at the function level. It preempts the question before a recruiter has to ask it.
Five cases spanning Alibaba ($50B+ GMV), Thermo Fisher (pharma supply chain, 0→1), Red Cross (mission-critical national deployment), Equinox+ (0→MVP in three months), and Allē (30M members, dual-surface). Recognizable companies, real scale, explicit role titles.
So why does this concern feel real to you even when the artifacts don't support it? I think I know. At Director+ level, the candidates you're competing against sometimes foreground organizational leadership artifacts: team growth narratives, hiring frameworks, design-system governance models, org charts showing reporting lines. Your cases foreground shipped product outcomes at leadership scale. That can feel IC-shaped when you're the one looking at it, because you're showing what the work produced rather than how you structured the organization that produced it. A hiring manager reading "secured executive buy-in," "led design and delivery for a national platform," and "0→1 build, six pharma partners, nine sites" sees someone who led and can prove what the leading produced. The organizational narrative belongs in conversation. The page is doing its job.
There is no /about or /bio page on the site. For your target roles, this is the right call. At this level, About pages are where candidates accidentally underlevel themselves by writing a chronological career narrative that buries the leadership signal under years of IC history. The homepage hero does the work in fewer words with stronger positioning. Some hiring committees do look for a career-arc page, but the ones evaluating for Head of Design are reading your case teasers and your essay, not a bio. If a specific search requires a narrative page, you can build one for that context. As a default, its absence is cleaner than its presence.
The problems sit in a different layer. And they are problems specifically because your thesis makes them problems.
P0 — Your private content is public
This is the highest-consequence artifact-level failure on your site.
Every full case study on junochen.com returns HTTP 200 with complete HTML to any direct request. Alibaba delivers 118KB of content. Red Cross delivers 248KB. The access control is a JavaScript check that looks for a localStorage key called pf_access_v1 and redirects unauthenticated visitors to a login page. But that check runs client-side, after the full document has already been served. Any recruiter using a link preview tool, any hiring manager whose browser has JavaScript disabled, any crawler, any curl request gets the entire case study without authentication.
The private pricing essay has the same exposure. HTTP 200, 34KB of HTML including the full article body, TinyFish pricing and launch-process material sitting behind a client-side password overlay. The overlay is cosmetic. The content is in the document.
Apply your own framework. You design systems where the wrong information reaching the wrong person at the wrong time creates real consequences. Your portfolio's access model is a locked glass door with the key taped to the frame. A hiring manager evaluating you for a role involving data governance, security-adjacent product decisions, or trust architecture will notice the gap between your thesis and your implementation. They may not mention it. They will note it.
Move full case content behind server-side authentication or a Vercel edge middleware check that returns 401/403 for unauthenticated requests. The teasers are public and strong. The full cases should require actual authentication, not a localStorage flag the server ignores. This is the single fix that eliminates the most common silent disqualification you'd never know about.
Verbal handle if it comes up before you fix it (use with caution — buys time, does not resolve the issue):
"The teasers are public by design. The full cases use a simple client-side gate because the audience is hiring managers I've already spoken with, not the open internet. I'm moving to server-side auth this month." Say it once. Don't elaborate. And actually do it.
The grain of truth here is real: you did build a gate, and the intended audience is pre-qualified. The problem is that the implementation doesn't enforce the intention. A confident senior operator can acknowledge that gap briefly. But this handle has a short shelf life. Every week it remains unfixed, the handle gets weaker.
P1 — Your Agentic inventory contradicts itself
The homepage Agentic section says "Three live agents built solo." The same page embeds four demo iframes: Carrier IQ, UAT Sentinel, Retail Velocity, and Brand Pulse. The JavaScript powering the "Now showing" interaction defines four entries. And there are five live Agentic subdomains, including Creator Scout, which doesn't appear on the homepage at all.
Three in the copy. Four on the page. Five reachable by URL.
Lower severity than the gating issue, but it hits the same nerve. A hiring manager evaluating you for a role involving AI product strategy will click through the Agentic section carefully. If the count doesn't match the inventory, and the inventory doesn't match what's reachable, the signal is: this person doesn't maintain consistency in their own published work. For someone whose thesis is trust through verification, that's an unforced error on the section most likely to differentiate you in AI-native searches.
Fix this week. Decide which agents are in the canonical set. Update the homepage copy to match. Either add Creator Scout to the page or take down the subdomain. The number in the copy must equal the number of demos on the page. Fifteen minutes.
Once the count is reconciled, there is no good verbal handle for the current state. If the page says three and shows four, any explanation sounds like you didn't notice. Fix it before your next conversation. After that, the handle is clean:
"The homepage shows the three production agents I built from concept to deployment. I have two more in testing that aren't featured yet." (High confidence, post-fix only.)
P2 — The missing Case Study 05
Your case numbering runs 01, 02, 03, 04, 06. No 05. The TinyFish URLs return 404, which means you've correctly removed the case from the site. But you left the numbering gap.
Minor in isolation. But apply the same lens you'd apply to any system you were evaluating: a removed record that leaves a visible gap in the sequence is a state-management failure. Something was taken out of the system, and the system wasn't updated to reflect the removal. Most visitors won't notice. The ones who do, particularly anyone evaluating your attention to system integrity, will see exactly the kind of inconsistency your own work is designed to prevent.
Fix when convenient. Renumber Allē to 05, or add a sixth case that fills the slot. Ten-minute change.
Verbal handle if someone asks (high confidence):
"That slot was a product case from my current role. I pulled it because the work is still in market and I want to respect the company's competitive position." True. Signals professional judgment. Makes the removal a feature.
P3 — Cached representations may still show "Product Designer"
The current server HTML is clean. But some crawlers and cached search results returned an older version of the homepage with the title "Juno Chen — Product Designer," a different H1, five Agentic cards, and a visible TinyFish case card. The server-side content is correct as of July 7. The problem is distribution lag.
This one is genuinely different from P0 through P2. Every issue above is an implementation failure on your site, something you control and haven't fixed. Here, your system is correct and third-party caches are distributing a version that no longer exists on your server. Worth noting because the distinction changes the response.
You cannot control what Google's cache shows today. You can accelerate the replacement.
Fix this week. Force a cache invalidation through Google Search Console. Resubmit the homepage URL for indexing. Check that your og:title and og:description tags (currently correct and leadership-forward) are being picked up by LinkedIn link previews, where recruiters are most likely to share or preview your URL. If LinkedIn still shows "Product Designer," delete and re-scrape the URL using LinkedIn's Post Inspector.
Verbal handle if a recruiter mentions seeing "Product Designer" (high confidence):
"I updated the site recently. You're likely seeing a cached version. The current homepage reflects my actual positioning: design leadership for high-stakes systems." Then send the direct link.
Priority stack
If you do nothing else before your next outreach:
-
Move full case content behind server-side auth. This is the only issue that could cause a silent disqualification you'd never know about. A security-conscious hiring manager, a CTO who inspects page source, a recruiter who runs your URL through a crawler. Any of them could encounter your "private" content served publicly and draw conclusions about your technical judgment that you never get the chance to address. Your teasers are strong enough to carry the public-facing portfolio. Let them.
-
Reconcile the Agentic inventory. Make the number in the copy match the number of demos on the page. Fifteen-minute fix. Eliminates an inconsistency in the section most likely to differentiate you in AI-native searches. No usable verbal handle exists until this is fixed.
Everything else is lower urgency. The case numbering gap is a ten-minute renumber. The cache drift will resolve within a week or two if you force re-indexing. The level signaling on the current homepage is already where it needs to be.
The trust surface you're being evaluated through
You built your professional thesis around a specific claim: trust in high-stakes systems depends on the system behaving consistently, transparently, and verifiably. Your portfolio is the first system a hiring manager evaluates you through. Right now, it says "private" and serves public. It says "three" and shows four. It numbers to six and skips five.
Fixable. All of it. And fixing them before they surface in conversation is always better than having a verbal handle ready for when they do. The verbal handles exist for the gap between now and when you ship the fixes. They are not a substitute for shipping.
The work itself is unambiguously Director+. The titles are explicit. The scope is real. The concern that your portfolio doesn't match your level is internal. You're comparing your shipped-outcome cases to a mental model of what "leadership portfolios" look like, and that mental model is wrong. Your cases show what leadership produced. The organizational narrative belongs in the room.
The best counter-narrative for "my portfolio doesn't match my level" is a portfolio whose own trust surface is airtight. Yours is close. Two fixes and it's there.
-
Backdoor references are rising: The Wall Street Journal reported that companies like Zapier now require live reference conversations before offers, with senior roles involving up to ten calls including off-book references, which means every verbal handle above must survive being repeated by someone who isn't you.
-
AI skills signal cuts both ways: A recent experimental study with 1,725 recruiters found AI skills boosted interview invitations by 8–15 percentage points overall, but effects were weaker for graphic designers due to recruiter skepticism toward AI in creative work, so your Agentic Labs section needs to demonstrate judgment, not just tool fluency.
-
Anthropic normalizes candidate AI use: Anthropic's published candidate guidance explicitly encourages using Claude for resume refinement, company research, and answer practice while reserving live interviews for unaided thinking, a distinction worth understanding before any conversation with companies that have similar policies.
-
Structured hiring reshapes the scorecard: Greenhouse's structured-hiring framework gives interviewers candidate scorecards tied to predefined skills and qualifications, which means your verbal handles need to map cleanly to evaluative criteria a panel member can write down, not just land well in conversation.