All articles
By Background··7 min read

AI Training Jobs for Mathematicians: 43 Live Roles, Pay and Requirements (2026)

43 live AI training roles for mathematicians from 5 platforms, with stated top rates from $65 to $110 an hour. What the work involves, who is hiring, what they ask for and what the pay figures mean.

Some links in this article are referral links — we may earn a small commission if you sign up, at no cost to you. This keeps the site free.

On October 11, 2026 our feed held 43 live listings for mathematicians, from 5 platforms. The number is a little misleading. 31 of the listings are fromMeridial, and 22 of those are for people who can do mathematics in a language other than English. Of the 12 listings from the other four platforms, 3 say outright that they are not tied to a current project.

This guide covers what the work involves, who is hiring, what they ask for and what the listings do and do not say about pay. Every figure comes from the listings themselves, and where a rate is the top of a range rather than an offer, we say so.

What AI companies want from mathematicians

Mathematical reasoning is a place where a fluent answer can be confidently wrong, so the work is mostly about checking. The listings describe six kinds of task.

Writing hard problems and grading proofs

Mercor's Mathematics Expert listing is the clearest description of the work. You might write a problem at research or competition level, with the statement, the intended solution and the reference answer a model's attempt is graded against. You might grade model-written proofs: whether each step follows, whether the edge cases are handled and whether an appeal to a known theorem actually satisfies its hypotheses. And you might flag where a model reaches a correct answer through an argument that does not prove it. The listing says most contracts in the field have paid $40 to $75 an hour, set per project by scope and depth.

Challenging a model and logging how it fails

Meridial's Mathematics Specialist roles, 24 listings in all, describe a more conversational version. A typical day, in the listing's words, is conversing with the model on classroom and theoretical problems, verifying accuracy and logical soundness, capturing reproducible error traces and suggesting improvements to prompts and evaluation metrics. The topics named include real and complex analysis, optimization theory, mathematical proofs, set theory, probability distributions and cryptographic algorithms.

Formal proof in Lean

The most specialized corner is formal mathematics. Mercor's Lean Engineer role ($90 to $110 an hour) asks you to write and review Lean 4 proofs against mathlib and to judge whether a formal statement is faithful to the informal one. It is a W-2 position with a minimum of 20 hours a week, rising to as many as 40. Meridial's Pure Mathematics Specialist listing asks for fluency in Lean 4 and an in-progress or freshly earned PhD, or postdoctoral experience.Handshake AI's listing for India names proof and formal methods (Lean, Coq, Isabelle) as one of three accepted sub-domains.

Computational and scientific tasks

Turing has two roles that mix mathematics with programming. The Scientific Computing Expert role for mathematics PhDs asks you to design terminal-based tasks in numerical analysis, optimization, statistics and dynamical systems, with solutions in Python, R, Julia, C/C++ or Bash. Its SciCode role asks for a Master's or PhD in mathematics and strong Python, and you write a main problem with at least three sub-problems, a verified solution and unit tests.

Quality review

The Mathematics Quality Assurance Lead role at SME Careers is for someone who oversees the quality, consistency and performance of other trainers, at up to $75 an hour. SME Careers is SuperAnnotate's expert network.

Geometry and computer vision

Meridial has 6 listings for homography and computer vision specialists. The work named includes projective geometry, multi-view consistency and labeling static points across images.

Who is hiring, and what they state

PlatformListingsWhat the listings stateNote
Meridial31$6 or $8 up to $65/hr (30); $35 to $65 for the Lean roleExact rate set by experience, expertise and location; 22 need a language other than English
Mercor3$40 to $75, $60 to $80 and $90 to $110/hrTwo are standing or open-application listings; the Lean role is W-2
SME Careers2Up to $70 and up to $75/hrThe Specialist role has no immediate project
Turing4$300 per approved task (1); no rate on the other 3The $300 role asks for 40+ hours a week for 5 weeks
Handshake AI3No rate on any listingThe India role asks for a PhD and first-author publications

What the pay figures mean

36 of the 43 listings state an hourly rate, and 31 of those are Meridial's. They share one ceiling of $65, which is the top of a stated range: Meridial's listings say they offer a pay range from $6 or $8 up to $65 an hour, with the exact rate determined after evaluating your experience, expertise and geographic location, and that final offers may vary. Treat $65 as the most you could be offered, not the usual rate.

The five hourly listings from other platforms state ceilings of $70 to $110, with a median of $75. Mercor's three also state floors of $40, $60 and $90. The highest stated hourly rate, $110, is for Mercor's Lean role.

What they ask for

Mathematics in other languages

22 of the Meridial listings are for mathematics specialists fluent in a second language: Arabic, Dutch, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Russian, Spanish and Swedish. There are 5 for Spanish alone. If you teach or do mathematics in one of these languages, these are the only mathematics listings in our feed that make a second language the stated requirement. Our bilingual speakers guide covers the language side.

Statisticians, actuaries and quants

Roles for statisticians sit under other tags in our feed. AfterQuery lists Statistician & Experimentation Expert, Biostatistics Expert, Actuary Expert and Econometrics Expert roles at $115 an hour, and a Quantitative Analyst Expert role at $175. Each of those shows a single figure in the feed, and AfterQuery sometimes stores the midpoint of a range, so the true top may be higher. Mercor's Data Scientist Talent Network lists $100 to $150. Our data scientists guide and finance guide cover them.

The catches

Common questions

Do I need a PhD?

Often, yes, but not always. Meridial calls a master's or PhD ideal in 24 of its 31 listings, Turing's roles ask for a master's or PhD and the Handshake AI listing for India asks for a PhD, while Mercor accepts an equivalent research or competition background. Our PhD guide covers the wider picture.

Do I need to know Lean?

Only for the formal-proof roles: Mercor's Lean Engineer and Meridial's Pure Mathematics Specialist. The others ask for written mathematical reasoning.

Which platform should I try first?

Mercor's Mathematics Expert listing is the most specific about the work and the pay range, but it is a standing listing. Meridial has the most listings and the only language-specific ones. Compare them on the pay report and read the platform pages for Mercor, Meridial and Turing.

How current is this?

This is a snapshot from October 11, 2026. Listings change daily. The live roles are on our mathematicians page.

Browse live listings →