AI Job Match reads your real work -- repos, papers, projects -- and watches every frontier lab's hiring picture. It surfaces the three to five roles a week you should actually look at. Quietly. No spam, no LinkedIn-ification, no telling the world you're looking.
Engineers in the network are matched to roles at
// The problem
You're not browsing a careers page. You're trying to figure out which three labs are doing work you'd actually want to do, who's hiring quietly, and whether your skill stack lines up with what the bar is right now. Most tools are tuned for the opposite of that.
// How it works
Set it up in three minutes. Then AI Job Match watches the labs and pings you only when something worth your attention shows up.
Link GitHub, Google Scholar, and any project write-ups or repos you want considered. Take 2 minutes to mark what you actually liked working on.
Pick the lab tiers, role families, geographies, and compensation floor you'd actually move for. Defaults are sensible; tune as you go.
Every night, AI Job Match crawls open and quiet roles across the labs we track and scores them against your profile. Only matches above your threshold get surfaced.
Want to talk? Hit "Open to introductions" on a match. Until you do, no lab and no recruiter sees you're considering anything.
// What it matches on
Six families of input the model uses to figure out what good fit actually looks like for you -- most of them ignored by every other job tool you've used.
Your public repos, what you've shipped, the size and shape of changes you've owned -- read structurally, not by name.
What you've authored or contributed to. Topic, depth, framing, and where the work sits relative to the frontier conversation.
Star ratings on past work, the "more of this" / "less of this" signals you give. It learns fast.
Pre-training research at a frontier lab is a different job than applied AI at a Series-A startup. The model knows the difference.
SF, London, Bangalore, fully remote, hybrid. AI Job Match honors hard constraints and treats soft ones as signal, not filters.
Network-verified offer data -- not headcount.com guesses. Tells you whether a role is at your bar before you spend an hour on it.
// Inside a match
No black box. Click any match and you see exactly why AI Job Match flagged it -- and which signals didn't quite line up. Honest scoring, not inflated.
SF / remote -- $880K-1.1M total -- This role opened to internal referrals 4 days ago and to the OpenTalent network this morning.
// Privacy & control
If you're currently employed and just curious, AI Job Match is built so that nobody -- labs, recruiters, or current employer -- sees you're looking until you flip the switch yourself.
Your profile is not visible to any lab or recruiter while in stealth mode. AI Job Match still runs, still surfaces matches, still tells you who's hiring quietly -- but nobody else knows you're considering anything.
Want to be open to one specific lab but invisible to everyone else? Toggle that lab. Per-lab visibility is the rule, not the exception. You can also block individual labs and the people who work there.
Labs and hiring managers in the network can only reach you after you've expressed interest in a specific match. No cold outreach. No "saw your profile" generic pings. Ever.
Your profile, your work links, your match history -- export it as JSON any time. Close your account and it's deleted within 30 days. Detailed retention is in the Privacy Policy.
// By the numbers
Active engineers using AI Job Match this quarter.
Match relevance score from engineer feedback on top-5 results.
High-quality matches per engineer per week, on average.
Of matches accepted become live conversations within two weeks.
I'd been ignoring my inbox for a year. AI Job Match was the first thing that surfaced a role I actually wanted to read -- and the reasoning panel told me exactly why. Took the call, took the offer.
Senior research engineer -- joined a frontier lab in Q1 2026
// FAQ
Yes, for members of the OpenTalent network. You can use AI Job Match as long as your network membership is active. If a match leads to a hire, the hiring lab pays OpenTalent. You never do.
Stealth is the default mode. Nobody -- not your current employer, not other labs, not recruiters -- has any way to see that you're considering opportunities until you explicitly flip "Open to introductions" on a specific match. You can also block any specific lab or any specific individual at any time.
No. We only read what you've explicitly connected, and only the data you've authorized via the GitHub OAuth scope you grant. You can revoke access at any time, and we delete the cached embeddings within 24 hours of revocation.
OpenTalent runs a quiet hiring channel for the labs in our network -- roles that aren't yet listed on careers pages or are open only to referrals. Members of our network see those roles. That's the channel AI Job Match draws from, alongside public listings.
No. Inbound is locked unless you opt in to a specific match. Labs and hiring managers in the network cannot cold-message you. That's a structural rule of the platform, not a preference setting.
Apply to the OpenTalent network. Less than 3% of applicants are accepted -- see the five-stage screening if you want a sense of the bar. Once you're in, AI Job Match is one click away.
// Other features in the network
AI Job Match is one of four features network members use to navigate the frontier AI job market.
AI Job Match is free for OpenTalent network members. Three minutes to set up. Stealth by default.