Tiger Tracks' "Top 1% in Cursor" JD: 34 Profiles Match on LinkedIn
A May 2026 HN job post replaced YoE with AI-tool fluency. LinkedIn surfaces 34 candidates. Here's how to source for it.
If you read the May 2026 "Ask HN: Who Is Hiring" thread, you noticed a JD line that didn't exist a year ago. Tiger Tracks, Inc. 5000 #123, ex-Google founders, is hiring a Founding AI Engineer and the required qualification is being "top 1% with Lovable, Replit, or Cursor" plus shipping a complex multi-API LLM tool in under a day. Years of experience isn't a filter. It isn't even mentioned.
This is the new gate, and your sourcing stack does not see it.
What the Tiger Tracks JD actually says
The exact text on the May 2026 thread: "top 1% with Lovable, Replit, or Cursor. Public GitHub or portfolio with shipped projects we can play with. Ability to ship a complex tool (multiple data sources, APIs, server hosting, LLM integration) in under a day."
The interview matches the JD. Thirty-minute intro, forty-five-minute founder conversation, then a sixty-minute live build in Lovable or Replit where they watch you ship. Offer within two weeks. The thing they want built is real: landing page conversion analyzers, video creative analyzers, vertical media buying co-pilots that touch Meta, Google, and TikTok APIs. This goes into client production the same week.
Tiger Tracks is not an outlier in that thread. Doubling is hiring a contract design engineer who ships code on a 0-to-1 mobile health product. Parabola wants an SRE for LLM-native ops tooling. HME Technologies posted an AI Ops Lead who builds with Make, Zapier, n8n plus Python and JS glue. Logenta's CTO listing gates on "agent harness, write clean typed Python, and have experience with tools like uv, DuckDB, and PydanticAI." Loophole Labs is filling its sixth founding seat on a K8s runtime.
The pattern is consistent. The qualification is demonstrated tool fluency and a public artifact. Tenure is invisible.
Why this isn't elitism, it's the death of credentialism
"Top 1% with Cursor" sounds like a flex. Read it again. Tiger Tracks is saying: we will hire anyone, from anywhere, with any title or no title, if they can ship a multi-API LLM tool in a day and we can play with what they've already built.
That replaces a much narrower filter. The old version of this JD said "5+ years at a top-tier tech company, CS degree, FAANG preferred." The new version says: open your GitHub. The credentialing class moved from Stanford and Google to the daily Lovable showcase feed and Replit Bounties leaderboards.
Years of experience went from gate to irrelevance in one JD line.
The HN dataset backs this up at scale. TrickCV's analysis of 6,539 tech postings from January 2025 through May 2026 found that 65% of roles labeled "Senior" now ask for five years of experience or fewer. Three hundred and twenty six postings, about 5% of the dataset, are explicitly "founding engineer" seats. YoE was already eroding before AI-tool fluency arrived to replace it.
The sourcing problem in one number
Here is where your pipeline breaks. The tools are huge. The signal on LinkedIn is not.
We ran the obvious structured query: AI Engineer or Founding Engineer title, Cursor as a listed skill, no geography filter. Thirty four people on the entire platform. For context, Cursor's maker Anysphere is at a $29.3 billion valuation with ARR growing from $1 million in late 2023 to over $1 billion by November 2025. Fifty thousand business customers. Millions of developer seats. The skill is real. The field is empty.
Lovable is worse. Four hundred million ARR in February 2026, 25 million total projects, more than 100,000 new projects launching every day. Search LinkedIn for "Lovable" as a skill and you will get noise about emotional intelligence. Replit hit roughly $253 million ARR by October 2025, growing 2,352% year over year, with non-technical teams at Coinbase, Zillow, and Mercedes-Benz building internal tools. None of that is on a profile.
This is what AI tool fluency recruiting actually looks like in 2026: a giant population, a structured-field blind spot, and a JD requirement that reads like a riddle if you only have Recruiter.
Why "vibe coder" undersells the candidate
The press tag for this hire is "vibe coder." That label is wrong for what Tiger Tracks is buying.
Karpathy, who coined the term, has already walked it back: "Vibe coding was exhilarating and fun as a local demo, but a bit of a painful slog as a deployed, real app." Researchers found that 170 out of 1,645 Lovable-generated apps, or 10.3%, shipped with critical row-level security flaws in their Supabase configurations. Accepting every diff and clicking publish is not what these JDs require.
What Tiger Tracks wants is an AI-augmented senior IC. Someone who uses Cursor Composer to move ten times faster than a 2022 engineer would, but who still has the architecture instincts to wire three APIs, an LLM, and a hosted backend together without leaking your customer data. The vibe coder sourcing problem is really an instinct-plus-velocity sourcing problem. Self-taught designers and ex-PMs absolutely qualify. So do laid-off staff engineers who learned Cursor on the side. Title is a poor proxy for either.
Kyrylai, a Toronto venture studio, ran a tidy receipt on this. Eight people on the Cursor Team Plan delivered one production-ready product, one semi-production tool, and three proofs of concept in ten weeks. They averaged 26.1 PRs per week with a 10.2-hour merge time. That is what "top 1% with Cursor" looks like in output. It is not a vibe. It is a throughput curve.
Where the candidates actually live
If LinkedIn skill tags are dead for this hire, you need a different map. The fluent population leaves traces in five places, none of which are on a resume.
GitHub portfolios with shipped artifacts. Not stars (we have written about the 6 million fake stars problem). Look for repos with deployed URLs in the README, recent commits, multi-API integrations, and a clean .cursorrules file. The "awesome-cursor-rules" repos are a useful starting graph for who is actually configuring the tool, not just installing it.
Lovable's daily showcase. A hundred thousand new projects a day, with a curated feed of the ones that ship. Authors are linkable. Most are not on LinkedIn under "AI Engineer."
Replit Bounties. Public completion history, public ratings, public code. It is the closest thing to a verified throughput leaderboard for the Replit engineer hiring pool.
Cursor Forum and the Cursor Discord. People who answer questions about Composer prompts and indexing strategies are, by definition, in the top 1%. None of them list Cursor on LinkedIn.
HN "Show HN" threads filtered for AI-builder tags. A founding AI engineer who posted a working tool on a Tuesday is a higher-signal lead than anyone in a Recruiter saved search.
Stitching those together by hand is the actual job. A single founding-engineer search can mean opening Lovable showcase, cross-referencing the author's GitHub, finding their X handle, confirming a real shipped project, and pulling a contactable email. Multiply by fifty leads.
This is the friction we built Refolk for. You describe the person in plain English ("top 1% Cursor builder with a deployed multi-API LLM project, ex-startup, US timezones") and Refolk searches GitHub, LinkedIn, and the open web together. The thirty four LinkedIn-skill profiles become a few hundred actually-fluent humans with shipped artifacts you can play with, which is what the Tiger Tracks JD asks you to bring on day one.
The new interview screens for the same thing you sourced for
The Tiger Tracks loop is a tell. A sixty-minute live build in Lovable or Replit is cheaper than a take-home, much harder to fake than a coding screen, and screens directly for the trait named in the JD. Expect this to spread to YC-style rounds within six months. The Y Combinator Winter 2025 batch already had 25% of startups running codebases that were 95% or more AI-generated. The interview will follow the codebase.
If you are a founder reading this: your job spec and your interview now need to match the actual production environment. If you are a recruiter: your sourcing has to surface people who can pass that live build, not people who can pass a 2021 phone screen. The two filters disagree on who is qualified, and the live build is going to win.
This is also why a tool like Refolk earns its keep on these searches. The pool is genuinely cross-pollinated: 63% of vibe-coding users are non-developers, which means designers, PMs, and solo founders are in your candidate set whether you wanted them or not. A LinkedIn title filter is actively harmful here because it removes the people most likely to pass the live build. A plain-English query ("shipped a customer-facing LLM tool in the last six months, comfortable in Cursor, no FAANG required") returns the actual pool.
A short checklist for the next two weeks
If you have an open founding AI engineer seat in May or June 2026, here is the order of operations.
- Rewrite the JD. Drop the YoE line. Name the tools (Cursor, Lovable, Replit, plus whatever your stack uses, Make and n8n if relevant). Ask for a public shipped artifact.
- Replace the take-home with a sixty-minute live build in the tool you actually use. Tiger Tracks is the template.
- Build your sourcing list from GitHub, Lovable showcase, Replit Bounties, the Cursor Forum, and HN "Show HN," not from a LinkedIn title search.
- Move fast. The HN thread refreshes monthly and other hirers are reading the same names off the same showcase feed you are.
The reason the Tiger Tracks JD is interesting isn't the "top 1%" flourish. It is that the gating signal moved off the resume, off LinkedIn, and onto artifacts that live in five different places on the open web. The hiring teams that learn to read those artifacts first will close. The ones still searching skill tags will keep finding the same thirty four people.
FAQ
How do I verify someone is actually "top 1% with Cursor" without watching them code?
Look at artifacts, not claims. A candidate's GitHub should show recent commits to a deployed project with a live URL, ideally one that integrates multiple APIs and an LLM. A clean .cursorrules file in the repo is a strong tell that they configure the tool rather than just installing it. Public Lovable showcase entries, Replit Bounties completions, and "Show HN" launches are higher signal than any self-rated skill. If you still aren't sure, do what Tiger Tracks does: a sixty-minute live build is the cheapest way to confirm.
Why do so few people list Cursor or Lovable as skills on LinkedIn?
Two reasons. First, the tools are too new and too informal for most engineers to treat as a credential, the way they would treat "Kubernetes" or "Go." Second, the fluent population includes a lot of non-traditional builders (designers, PMs, solo founders) who do not maintain a LinkedIn skills section at all. Our index check returned about 34 profiles globally with "Cursor" listed alongside an AI or Founding Engineer title, against a real population in the millions.
Is "vibe coder" the right title to search for?
No. The label is catchy but misleading for serious hiring. The Tiger Tracks JD and most adjacent May 2026 postings demand shipping complex multi-API tools with real architecture judgment, not "accept all diffs" prompting. Search for the artifact (a shipped LLM tool with public code) and the tooling (Cursor, Lovable, Replit, sometimes plus uv, DuckDB, PydanticAI). The title will sort itself out.
What does this mean for traditional senior engineers with ten years of experience?
They are still very hireable, but only if their recent work shows AI tool fluency in artifacts a hiring manager can play with. A 2021 staff engineer resume with no public shipped LLM project will lose this seat to a 2024 self-taught builder who has three deployed tools on Lovable. The fix is straightforward: ship something public this quarter, using the tool the JD names, and link it from the top of your GitHub.