The PSE Wind-Down Just Freed ~49 ZK Engineers. None Will Post on LinkedIn.
The Ethereum Foundation cut 54 on June 23, 2026 and wound down PSE. Here is how to source the ZK cryptographers before Ethlabs and Aztec grab them.
On June 23, 2026, the Ethereum Foundation eliminated 54 roles, cut its 2026 budget by 40%, and wound down Privacy & Scaling Explorations, its applied ZK research unit. One day earlier, five senior EF researchers had already announced Ethlabs, an $11B-backed independent lab. If you are hiring ZK cryptographers or protocol engineers this quarter, the pool just moved, and it did not move to LinkedIn.
What actually happened on June 23
The headline number is 54 positions, roughly 20% of the Foundation's ~270-person staff. The subtler number is the restructure into five domain clusters and the spending glide path from about 15% of treasury today toward a 5% target by 2030. That is not a one-time correction. That is a permanent capacity reduction inside the single largest concentration of applied ZK talent in crypto.
The PSE wind-down matters more than the headcount. PSE shipped Semaphore (anonymous signaling), MACI (private voting), zkEmail, zkTLS, and Anon Aadhaar. Those repos are how the ecosystem got usable zero-knowledge primitives out of academic papers and into apps like clr.fund, Gitcoin's Allo stack, and Optimism RPGF tooling. When the org that maintains them shrinks, the maintainers scatter.
Severance is at least one month per year of service, plus transition grants and, notably, "ecosystem placement assistance." Vitalik Buterin explicitly wished departing employees would keep contributing to the "wider Ethereum ecosystem, or even the wider CROPS world." That last phrase is the tell. This is a warm handoff, not a market dump. If you wait for these engineers to update a headline, someone with a Telegram DM and a signed term sheet will have moved first.
Why LinkedIn will miss almost all of them
The Ethereum Foundation is a Swiss Stiftung. There is no WARN filing. There is no US state disclosure. There is no 8-K. The severance terms were disclosed in a blog post and repeated across thirdweb, TechTimes, and The Defiant, not in a regulatory system your ATS knows how to ingest.
More importantly, this cohort does not maintain LinkedIn profiles the way a Series C SaaS engineer does. Look at PSE's own release notes: Semaphore V2 credits cedoor and akinovak as core contributors. The MACI coordinator repo lists a security contact at paik@a30a.dev. These are pseudonymous or semi-pseudonymous GitHub identities with years of commits, PR reviews, and EIP co-authorship, and often no public employer field.
The Boolean strings that work for a React engineer do not work here. "Zero-knowledge" as a LinkedIn title match returns crypto-marketing hires. "Circom" returns 3 people, most of them mislabeled. Internal professional-network data we have seen puts strict "ZK + Circom + Solidity" applied cryptographers at roughly 15 profiles globally. Fifteen. The PSE wind-down alone likely doubles the reachable pool, but only if you know where to look.
The sourcing surfaces that actually work
Four places to start, in priority order:
github.com/privacy-scaling-explorationsand its sub-orgs. Pull PR authors and reviewers from the last 24 months. Cross-reference against release-note credits.github.com/semaphore-protocol. The main Semaphore repo has 1,063 stars and 293 forks. The commit graph names the maintainers who actually ship.github.com/zkemailandgithub.com/anon-aadhaar. Smaller repos, tighter contributor lists, higher signal per handle.- EIP authorship on
ethereum/EIPs. Julian Ma co-authored EIP-7805 (FOCIL) for censorship resistance. Ansgar Dietrichs is central to proposer-builder separation work. Barnabé Monnot led robust incentives research. Those names are trivially reachable from the EIPs repo, but so are the second-tier co-authors most recruiters skip.
Devcon speaker archives (Bogotá 2022, SEA 2024) are the fifth surface, and the most time-limited. The Foundation itself has telegraphed that Devcon will get smaller and cheaper to run. Future editions will surface fewer new names, so the historical lists compound in value.
The five you cannot hire, and the ~49 you can
Press coverage has fixated on the Ethlabs founding team: Ansgar Dietrichs, Barnabé Monnot, Caspar Schwarz-Schilling, Josh Rudolf, and Julian Ma. They spent years at the Foundation on the protocols that govern how Ethereum reaches consensus, processes transactions, and resists manipulation. Ethlabs is backed by Bitmine (which holds ~5.67M ETH, roughly 4.7% of supply, worth about $9.8B), SharpLink (872,984 ETH, ~$1.5B as of May 11, 2026), Joe Lubin, Anchorage Digital, Octant, and SNZ. Those five are not on the market. Do not waste an intro request.
The other ~49 are. So are the senior departures from the preceding six months: co-executive directors Tomasz Stanczak (February) and Hsiao-Wei Wang (June 18), protocol coordinators Tim Beiko and Barnabé Monnot (May), plus Josh Stark, Trent Van Epps, Carl Beekhuizen (seven years at the Foundation, co-designed the Beacon Chain), and Julian Ma. Nine senior names since January, most of them EIP-authoring, Devcon-speaking, and searchable by GitHub handle.
The five Ethlabs founders are a distraction. The applied cryptographers shipping MACI and zkEmail day to day are the actual hiring event.
The critical question, and the one every ZK-rollup CTO is asking privately, is whether applied engineering capacity for privacy work will exist outside the Foundation in orgs like Ethlabs or in the broader open-source community. If you are staffing that capacity, you are competing with a very short list of concrete buyers: Lagrange Labs, Succinct, Aztec, Gensyn, StoffelMPC, Tokamak Network, Primefactor, and OKX. That is where the diaspora will cluster. That is who you are bidding against.
The plain-English query that beats a Boolean string
The reason recruiters lose these searches is not effort. It is that the query language does not fit the target. "Applied ZK cryptographer, ex-PSE, contributes to Semaphore or MACI, has co-authored an EIP, based in Europe or LATAM, open to a seed-stage founding engineer role" is a sentence. It is not a Boolean. And it maps to zero LinkedIn filters cleanly.
That mismatch is why we built Refolk: you describe the person the way you would describe them to a co-founder, and you get a ranked shortlist that spans GitHub, LinkedIn, and the open web in one pass. GitHub commit graphs, EIP co-authorship, Devcon talks, and pseudonymous handles all fold into the same result set. You do not have to pick which surface to search first.
Two follow-on queries that work on the same shape:
- "Contributors to zkEmail or Anon Aadhaar who also have Circom experience, currently listed at a non-crypto employer or with no employer field."
- "Beacon Chain contributors from the 2020 to 2022 window who have not posted a job update since March 2026."
Both of these are impossible to type into LinkedIn Recruiter. Both are one sentence in Refolk.
The AI-privacy angle nobody is pricing in
Here is the trade most crypto recruiters are missing. The same ZK and MPC skillset that shipped MACI is the skillset now powering zkML, private inference, and AI-agent identity. The Foundation's new Protocol Cluster mandate keeps "L1 privacy" as a research goal, but the applied execution capacity is being disbanded exactly as AI founders start budgeting for private inference.
If you are an AI infra founder, you are not competing with Aztec for these hires. You are competing with a version of Aztec you have not modeled yet. The engineers who built Semaphore can build a private-inference proving pipeline. Very few people can. The ones who can are, this quarter, unusually reachable.
This is where a plain-English sourcing layer earns its keep twice. The same Refolk query that surfaces PSE contributors for a rollup team surfaces them for an AI startup, because the underlying signal (GitHub commits to a Circom circuit, a talk at zkSummit, a co-authored EIP) is orthogonal to the JD's industry label. You describe the human. The industry filter is your problem, not the sourcing tool's.
A concrete 10-day plan
If you want to move before the "ecosystem placement" pipeline routes people to the obvious homes, the window is measured in weeks. Here is the sequence that works:
Days 1 to 2. Pull contributor graphs from privacy-scaling-explorations, semaphore-protocol, zkemail, and anon-aadhaar. Rank by merged PRs in the last 18 months. Deduplicate against the Ethlabs five and the known senior departures already spoken for.
Days 3 to 4. Cross-reference against EIP co-authorship on ethereum/EIPs. Weight for FOCIL, PBS, and MEV-related EIPs, but do not skip smaller EIPs (those authors are less-covered and more reachable).
Days 5 to 6. Cross-reference against Devcon Bogotá 2022 and SEA 2024 speaker lists. Anyone who spoke on a PSE-adjacent topic and is not already at Ethlabs is your inbound priority.
Days 7 to 8. Widen to downstream integrators: clr.fund, Gitcoin (Allo stack), Optimism RPGF, EthMexico. These are secondary PSE-fluent devs, not primary cuts, but they are the closest bench.
Days 9 to 10. Reach out. Not with a JD. With a specific reference to a PR they merged or an EIP they co-authored. This cohort filters ruthlessly on whether the sender read the code.
If you are running that sequence across two roles, you can do it in spreadsheets. If you are running it across a team of six recruiters and eight JDs, Refolk collapses the first eight days into a query. That is the honest pitch.
What "ecosystem placement" means for your close rate
The Foundation is not sending resumes to a job board. "Ecosystem placement assistance" in Swiss-severance context means warm introductions from EF leadership to the orgs EF leadership already trusts: Ethlabs, Aztec, the client teams, and a handful of L2s. If your logo is not on that list, you are not in the default routing.
That is not a reason to give up. It is a reason to move first and move specifically. A founding engineer offer at a well-funded AI-privacy startup, delivered before the Foundation's internal routing catalog assigns the candidate to a rollup team, wins more often than the market thinks. The reason it wins is that the offer is legible in the candidate's own language: a specific problem, a specific repo, a specific co-author.
The PSE wind-down is a rare event. There will not be another concentration of applied ZK talent this reachable this cheaply for at least a cycle. The recruiters and founders who treat it as a normal layoff will source it like one, and lose. The ones who treat it as a GitHub-graph problem with an eight-week clock will not.
FAQ
How is this different from a normal tech layoff for sourcing purposes?
The Ethereum Foundation is a Swiss Stiftung, so there is no WARN filing and no US disclosure. There is no LinkedIn signal, because many of the affected engineers use pseudonymous GitHub handles and do not maintain resumes. And severance explicitly includes "ecosystem placement assistance," which means the Foundation is warm-handing candidates to a short list of preferred orgs. If you are not on that list, the only way in is to reach the candidate directly through GitHub or EIP co-authorship signals before the internal routing catches up.
Should I try to hire from Ethlabs?
No. The five Ethlabs founders (Ansgar Dietrichs, Barnabé Monnot, Caspar Schwarz-Schilling, Josh Rudolf, Julian Ma) just launched a lab backed by Bitmine, SharpLink, Joe Lubin, Anchorage Digital, Octant, and SNZ, with roughly $11B of ETH treasury exposure behind it. They are not moving. The hiring opportunity is the ~49 other cuts and the nine senior EF departures from the preceding six months, most of whom are not partners at an $11B org and are actively looking.
What GitHub orgs should I actually be pulling contributor lists from?
Start with privacy-scaling-explorations, semaphore-protocol, zkemail, and anon-aadhaar. Then cross-reference against ethereum/EIPs for co-authorship, weighting FOCIL (EIP-7805), PBS, and MEV-related proposals. Then widen to downstream integrators (clr.fund, Gitcoin's Allo stack, Optimism RPGF tooling) for adjacent PSE-fluent devs. Release-note credits inside those repos, following the cedoor and akinovak pattern from Semaphore V2, are your highest-signal shortlist.
Is this only useful for crypto companies?
No, and this is the trade most people are missing. The same ZK and MPC skillset that shipped MACI and Semaphore powers zkML, private inference, and AI-agent identity work. AI infra founders should be sourcing the PSE diaspora as aggressively as ZK-rollup teams are. The pool is small (roughly 15 global profiles on a strict Circom-plus-Solidity filter before the wind-down) and the overlap between "can build a Semaphore circuit" and "can build a private-inference proving pipeline" is very high.