AI is sourcing tomorrow's candidates. Are you on its radar?
Tomorrow is already today.
Paris, FR · French · English
Camille Martin
CTO & co-fondatrice @ Hyperline · auteur de stack-router
Currently at
Hyperline (YC W25)
CTO & Co-founder · 2024–Present
Building Hyperline
Paris, FR · ouvert advisor / collab OSS
About
CTO et co-fondatrice de Hyperline (YC W25), plateforme de pricing programmatique pour SaaS B2B. Auteur de stack-router (3,8k ★, 1,2M dl/sem npm), sponsorisée par Cloudflare. 7 ans en backend distribué avant la fondation.
Skills
2024 – Present
CTO & Co-founder
Hyperline (YC W25)
Paris, FR
Co-fondatrice et CTO de Hyperline, plateforme de pricing programmatique B2B. Architecture event-driven sur GCP. Levée seed 2,1M€ (Index, Kima). Équipe technique de 6, recrutement en cours.
2019 – 2024
Senior Backend Engineer
Qonto
Paris, FR
Senior sur l'infrastructure paiements. Designed et shipped le service de comptes pro pour 500k clients SMB. Pilotage de la migration AWS → GCP de l'équipe banking core.
2017 – 2019
Backend Engineer
Algolia
Paris, FR
Travail sur le moteur de search distribué. Optimisation latence P99 sur les régions EU, contribution au sharding multi-tenant.
Education
M.Sc.
Computer Science
EPITA
2017
Project · 2024
TypeScript · Rust · Edgestack-router
Librairie open-source de feature-flagging avec routing edge. 3,8k ★ GitHub, 1,2M dl/semaine npm. Sponsorisée par Cloudflare depuis 2025.
Project · 2023
Postgres · Rustqonto-pg-bench
Suite de benchmarks Postgres orientée fintech (latence sur les workloads de paiement). Utilisée en interne avant open-sourcing partiel.
Contact
inlinkedin.com/in/camille-martin-fr
@••••@•••.com
ghgithub.com/camille-martin
Native indexing supported by
Built on the llms.txt spec.
How it works
3 steps to maximize your chances when an AI agent searches for someone like you.
Upload your CV. We turn it into a structured version LLMs love to read — dated, hashed, controlled by you.
Upload & anchor
Drop your LinkedIn PDF. We hash it on upload — the cryptographic fingerprint anchors your profile to a real document at a real moment. Edit later as you wish; the original source stays verifiable.
linkedin_export_2026.pdf
sha256:eb3cc1a7…d93e8c
Extract & structure
Our LLM reads the PDF and emits your structured llms.txt layer and Bento grid simultaneously — no free-form rewrites.
name:Camille Martin
headline:CTO @ Hyperline
skills:TypeScript,Rust,K8s
role:CTO@Hyperline (YC W25)|2024–Now
llms-full:/camille-martin/paris/llms-full.txt
Publish
Ship one URL. Three endpoints: a Bento for humans, a public llms.txt, and a gated llms-full.txt for verified recruiters.
/Bento · humans/llms.txtpublic/llms-full.txtgated · TollBitOne file in. One URL out. Read by every human and every agent.
More and more, AIs are the ones finding candidates for recruiters.
Their agents, ChatGPT, Claude, Gemini — they all search through what they can index publicly: personal sites, github.io, blogs, Malt. If your profile isn't there — or is outdated — they don't find you, or worse, they make things up. Open-Shadow publishes a machine-readable summary of your CV: up-to-date, hashed, controlled by you.
Tomorrow · a typical recruiter query
“Find me a senior React/Django dev in Nantes for a Tech Lead role.”
Théo Bernard
source: theobernard.dev
Tech Lead React/Django · active blog
Marie Lefèvre
source: marielefevre.io
Full-stack React/Django · 8 years (self-declared)
Julien Faure
source: github.io/jfaure
React + Django (mentioned in GitHub bio)
LLMs only search through what is publicly indexable: personal sites, github.io, blogs, Malt. Without a web presence, you never appear — regardless of your actual level.
Yanis Moreau
92/100Tech Lead @ ManoMano · React · Django · K8s
React/Django Lead in Nantes, 7 years, aligned stack.
Lucas Petit
87/100Senior Backend @ ManoMano · Django · GraphQL · Postgres
Senior Django + 2 years as Lead, solid foundation.
Émilie Roux
84/100Senior Full-Stack @ MyMoney · React · Django · AWS
React/Django in fintech, product context.
5 cryptographically verified candidates, up-to-date, ranked in 800ms. No personal site needed: your llms.txt is enough.
For a recruiter to find you via ChatGPT, you have to be in the index the AI queries. Today it's the public web — biased toward the most visible candidates. Tomorrow, it'll be our endpoint — biased toward actual fit.
Enter the indexFor an LLM to find the right profiles, it needs another source to read besides the public web.
Open-Shadow only becomes the source LLMs query if enough of us publish our profiles to it. The more of us there are, the less they need to scrape or guess. Help us become the first source LLMs query for professional talent.
Before you claim your slug.
The real questions — on data, privacy, legal, and price.
What is llms.txt?+
A plain-text file that describes you in a format AI agents can read instantly — no PDF parsing, no HTML scraping. It looks like this:
name:Camille Martin
headline:CTO @ Hyperline
skills:TypeScript,Rust,K8s
Think of it as your robots.txt, but for your career.
Is it free?+
Yes. Upload your PDF, get your profile, share your link — all free. Recruiters pay micro-fees via TollBit when their agents access the premium llms-full.txt layer. Candidates never pay.
Where is my data stored? Can I delete it?+
All profiles live in an EU-region Supabase instance (Frankfurt). Your data is yours — export or delete any time from your dashboard. Deletion is full: profile row, source PDF, and hash chain removed. GDPR-compliant by design, not retrofit.
Is my contact info public?+
You control every field individually. Each contact field can be set to one of three modes:
Phone, email, and custom links each have their own toggle. Nothing is exposed unless you opt in.
Can I stay anonymous?+
Yes. Shadow Mode hides your name, photo, and contact info on the public Bento page. Your llms.txt is served as name:anonymous. Only verified recruiter agents with a TollBit token can access your routing data in the gated layer.
What if my name is common? Can someone else take my slug?+
Slugs are city-scoped: /jordan-t-smith/new-york and /jordan-t-smith/paris coexist. Within a single city, first-claim wins, but collisions are resolved by matching the source PDF hash and LinkedIn URL. Claim early if you're worried.
How do AI agents actually find profiles?+
Every profile publishes a llms.txt at a predictable URL. Agents discover profiles two ways:
- Direct crawl — the agent fetches /name/city/llms.txt
- Semantic search — the API matches a query against pgvector embeddings and returns ranked profiles
Both return pure structured data — zero HTML to parse.
Will ChatGPT or Gemini suddenly know about me after I upload?+
Not exactly — and understanding why is the whole point. Here's what actually happens:
- Direct fetch — When a recruiter's agent fetches your URL directly, it gets your verified profile instantly. That's the surface you control.
- Cold query — When a recruiter asks ChatGPT cold ("who is Marie Dupont?"), the model answers from training data — which won't include you for months, even years. That's the gap your side-by-side proves.
- Live retrieval — Increasingly, recruiter tools chain LLM + live web fetch — they query the model and fetch your llms.txt at the same time, then merge. As more agents adopt llms.txt, more answers get grounded in your data.
The goal isn't that AI suddenly knows you. The goal is that when an agent does look you up, your truth wins — not the model's hallucination.
What does this cost my team?+
The public llms.txt layer is free — any agent can read it. The gated llms-full.txt is metered via TollBit at roughly $0.002 per crawl, with volume discounts above 100k reads/month. No seat licenses, no minimums. Pay only when your agents read.
What does the Shadow-Badge guarantee?+
It means the profile has cryptographic provenance — anchored to a real PDF the candidate uploaded. Specifically:
- ✓ Profile anchored to a real source PDF via cryptographic hash
- ✓ Source hash is stored — proof the document was real at a real moment
- ✓ Faithful extraction at parse time — no LLM rewrites; later edits are explicitly authored by the candidate
Agents see the full provenance trail — and the candidate's edits are clearly theirs, never AI hallucination.
Won't LinkedIn block or sue you for this?+
No. Open-Shadow never scrapes LinkedIn. You initiate the export from LinkedIn yourself and upload the PDF — under hiQ Labs v. LinkedIn (9th Cir. 2022) that's your data to share. We don't query LinkedIn APIs, reverse-engineer, or proxy. The wedge is that you own the document.
Still missing something? email us.
Transform your CV.
Join the AI index.
Convert your PDF into a format native to LLMs. A dated, cryptographic source of truth — fully under your control. No account required.