587 vs 53,853: Welsh Was Right About Rust. Source, Don't Post.

Matt Welsh's "Revisiting Rust in 2026," published March 15, 2026, told founders what recruiters already knew: post a Rust role and you will drown in JavaScript, Python, and C++ resumes "by orders of magnitude." The September 2026 Hacker News "Who is hiring?" thread ran roughly 260 posts, a handful of them Rust-first (Rerun, Estuary, Kepler AI), and the pattern was identical across every one. The fix is not a better job description. The fix is to stop posting and start naming people.

The gap Welsh described is literal, not rhetorical

In Refolk's index, there are 587 US-based engineers who list Rust as a skill and hold a core IC engineering title (Software, Backend, Systems, or Staff Engineer). The same query for Python returns 53,853. That is a 1:92 ratio, and it is the mechanical reason every Rust posting fills up with Python resumes.

Welsh's TL;DR from the essay: "adopting Rust is still a moderate (if not exceptionally strong) headwind for moving quickly, and moving quickly is essential for early startups." He carves out robotics and safety-critical embedded, which is exactly where the September HN Rust postings cluster: Physical AI (Rerun), data infrastructure (Estuary), autonomous systems (Welsh's own current team at Palantir in Seattle).

The mechanism matters. Rust adoption in the US concentrates at a small number of employers. Oxide Computer appears twice in Refolk's top-10 US Rust IC employers despite being a small company, alongside Google (3) and Meta (3). The pool never diffuses into the general job market, so an inbound funnel samples the base rate of engineers, not the base rate of Rust engineers. There is no keyword you can add to a job spec that turns 53,853 Python engineers into 587 Rust ones.

92x

What the September 2026 HN thread actually contained

The September 2026 "Who is hiring?" thread indexed roughly 260 posts as of September 15, and the Rust-first employers within it were a tight, named list. That is your total addressable demand for the month, and it fits on an index card.

A few employers are worth naming because they tell you where the pool actually lives:

• Rerun (Stockholm with an SF hub). Physical AI data stack. Most of the core, performance-critical code is Rust. Co-founder Emil Ernerfeldt created egui, one of the most widely used Rust GUI frameworks. The person who wrote your dependency is running the company that is hiring.
• Estuary. Data pipelines. The June 2026 HN post read "Strong backend fundamentals required; Go and Rust experience preferred." Rust is a preference, not a filter.
• Kepler AI (NYC, founded by ex-Palantir Foundry engineers). Rust backend. Explicit line in their post: "We don't require Rust experience; we hire on fundamentals."
• Oxide Computer Company. Two seats in Refolk's top-10 US employers of Rust ICs. A single Rust-native shop absorbs a share of the pool that any FAANG would kill for.

Kepler's line is the leading indicator. When Rust-first shops publicly drop the Rust requirement, the shortage has moved past "hard to fill" and into "restructure the job." That is the shape of the 2026 Rust job market.

The numbers, side by side

Two numbers do the heavy lifting. The 587 is the entire US IC Rust pool worth talking to. The 32% Berlin concentration in Germany is the punchline of the next section.

Berlin's Rust pool is a founder pool, not a hire pool

In Refolk's German Rust sample, Co-Founder and CTO titles outnumber Senior Software Engineer roughly 8:2 among the top titles. Germany has 967 Rust profiles total, and the ones concentrated in Berlin are largely running their own companies, not looking for a job. Recruiters who "source Berlin Rust" are usually pitching to their targets' own cap tables.

That changes the outreach entirely. A standard "we're hiring senior Rust engineers, would love to chat" note to a Berlin CTO is dead on arrival. What works, in rough order of hit rate:

1. Acqui-hire framing. If the target is a two-person startup, the honest conversation is about buying the team, not recruiting an IC.
2. Advisory or fractional CTO. Founders with real Rust chops will trade a day a week for equity and a strategic problem.
3. Contract-to-perm on a specific hard problem. Not a role. A named piece of work with a deadline.
4. Recruit their next hire, not them. A founder who likes your company will introduce you to the last three Rust engineers they tried to hire and lost.

The same pattern is emerging in Stockholm around Rerun and in San Francisco around Oxide alumni. Rust talent clusters, and the cluster is often incorporated.

Where Rust engineers actually are: GitHub, crates.io, conference stages

The real sourcing surface for Rust is GitHub contribution graphs, crates.io maintainer lists, and RustConf speaker rosters. LinkedIn boolean search misses the people who matter, because the people who matter identify by project, not by keyword. Emil Ernerfeldt would never surface on a search for "Rust engineer"; he surfaces as the author of egui.

If you want a working list this week, start here:

• Top contributors to tokio, axum, hyper, tonic. The async runtime and web stack. Almost anyone in the commit graph is hireable as a senior backend engineer.
• Maintainers of polars, datafusion, arrow-rs. Data infra Rust. Rerun, Estuary, and every "we rewrote it in Rust for the perf" startup wants these people.
• burn, candle, tch-rs contributors. Rust ML. Small pool, high leverage, most already know each other.
• egui, iced, dioxus contributors. Rust UI. Ernerfeldt is the proof of concept: build the framework, hire from the users.
• RustConf 2025 and 2026 speakers. Public list, public talks, easy to reach. Half of them will not appear in any recruiter database.
• The Rust project's own governance teams. Compiler, libs, infra. Overrepresented at Oxide, Meta, AWS, and Ferrous Systems.

This is exactly the work that breaks a keyword tool. You do not want "Rust engineer in the Bay Area." You want "people in the tokio commit graph in the last 18 months who are not currently at Oxide or AWS." That kind of query is why Refolk exists: you describe the person in plain English, across GitHub, LinkedIn, and the open web, and get a ranked shortlist instead of a keyword dump.

The convertible-candidate strategy

The second-best Rust sourcing list is not Rust engineers. It is strong C++ and Go engineers with correctness-critical backgrounds who can be converted, which is exactly what Kepler AI is publicly doing. Welsh's own essay concedes that LLMs "have changed the game in both good and bad ways for adopting a new language like Rust," and Cursor plus Claude Code makes the ramp meaningfully shorter than it was in 2022.

The convertible profile looks like this:

• Five plus years of C++ in a memory-safety-sensitive domain (databases, browsers, kernels, HFT, robotics).
• Or Go at a company that took correctness seriously (Cloudflare, CockroachDB, Tailscale, Sourcegraph).
• Public evidence of caring about types, lifetimes, or performance. Blog posts, conference talks, or a personal project in Rust that never went anywhere.
• Currently at a company that has slowed down or reorganized. Reorg is the single strongest "will take a call" signal in 2026.

The pitch is not "we need a Rust expert." The pitch is "we hire on fundamentals, our stack is Rust, and you will be productive in six weeks with Claude in your editor." Kepler ships that pitch in public. Rerun and Estuary ship softer versions of it. The recruiters winning Rust hires in 2026 are the ones who built a convertible list twice the size of their pure-Rust list, and worked both in parallel.

The 587 pure-Rust ICs are the seed list. The convertible list is where the hires actually close.

How to build the outbound list in a week

The concrete workflow, if you have one requisition and seven days:

1. Day 1: enumerate the pure-Rust pool. Pull the named list of US Rust ICs. It is 587 nationally, heavily concentrated in the Bay Area (6 of a 25-person sample). Cross-reference against current employer to remove people at Oxide, AWS Rust teams, and Meta systems, who almost never move.
2. Day 2: layer in GitHub signal. Rank the remainder by contributions to tokio, axum, polars, burn, and the Rust compiler in the last 18 months. Describing "engineers whose GitHub shows sustained Rust contribution but whose LinkedIn does not headline it" is the exact kind of ask Refolk is built for.
3. Day 3: add RustConf and Rust Nation speakers from 2024 through 2026. Public rosters. Deduplicate.
4. Day 4: build the convertible list. C++ engineers at databases and browsers, Go engineers at correctness-first infra shops, robotics C++ engineers displaced by 2026 layoffs.
5. Day 5 to 7: outreach. Two tracks, two message templates. The pure-Rust track leads with the specific project you want them to own. The convertible track leads with "we hire on fundamentals" and names a mentor on the team.

If you run this weekly for a quarter, you will have talked to a meaningful fraction of the entire US Rust IC pool. That is only possible because the pool is 587 people. It is also the reason inbound cannot work: 587 people are not reading your job board.

What Welsh's essay actually means for recruiters

Welsh's argument is that Rust is a headwind for startup velocity except in a narrow band of domains, and that LLMs have changed the calculus in both directions. Translated into recruiting terms: fewer companies will insist on Rust experience going forward, more will hire generalists and train them, and the pure-Rust pool will stay small because it is not growing at anything like the rate of Rust adoption on Hacker News.

The recruiter takeaways:

• Stop measuring Rust roles by applicant volume. The right metric is named outbound sent per week.
• Budget for a dual pipeline: pure-Rust for the roles that truly require it, convertible for everything else.
• Treat the roughly 260-post HN thread as a competitive map, not a candidate pool. The employers in it are your comp set for compensation, mission, and stack.
• Read the 2025 State of Rust Survey (published March 2, 2026) for the adoption-vs-hiring gap in your specific domain before you write the spec.

Welsh closed his essay hiring at Palantir in Seattle for autonomous systems. He is competing for the same 587 people you are. The recruiters who win in 2026 are the ones who accept the number and act like it.

FAQ

How many Rust engineers are actually on the US job market in 2026?

Refolk's index shows 587 US-based engineers who list Rust as a skill and hold a core IC engineering title (Software, Backend, Systems, or Staff). That is the ceiling on any pure-Rust sourcing campaign in the US, and the pool is heavily concentrated at Oxide Computer, Meta, Google, and a small tail of Rust-native startups. The addressable subset (people actually open to a move) is smaller still, which is why outbound to named individuals beats any inbound strategy at this pool size.

Why do Rust job postings fill up with Python and JavaScript applicants?

Because the base rate of Python engineers is roughly 92 times higher than Rust in the US IC pool, so even a Rust-specific keyword filter cannot overcome the noise floor. Matt Welsh's March 2026 essay called this "orders of magnitude," and Refolk's index confirms the multiple is literal. The mechanism is that Rust adoption clusters at a small number of employers rather than diffusing into the general job market, so postings sample the general market and miss the Rust cluster entirely.

Should I require Rust experience or hire on fundamentals?

Kepler AI and Estuary are publicly hiring on fundamentals, and the shortage justifies it. The convertible profile is a C++ or Go engineer from a correctness-critical background with public evidence of caring about types and performance, who can ramp to production Rust in roughly six weeks with Claude or Cursor in the editor. Keep the Rust requirement only for roles where the first thirty days matter more than the first year, such as a solo Rust hire on an existing team with no mentor bandwidth.

Where do I find Rust engineers who do not show up on LinkedIn?

Start with GitHub contribution graphs for tokio, axum, polars, burn, and the Rust compiler itself, then layer in crates.io maintainers of any crate with meaningful download volume and speaker rosters from RustConf and Rust Nation 2024 through 2026. Many of the strongest Rust engineers headline something else on LinkedIn (previous job, previous language) and are only visible through their open-source work, which is why plain-English sourcing across GitHub and the open web finds people that boolean keyword search on LinkedIn structurally cannot.