Cerebras

Cerebras

cerebras.net
AI-native81 open roles

Wafer-scale AI chips for training and inference.

Signals updated

Cerebras is an AI-native company — Wafer-scale AI chips for training and inference. Our index currently tracks 81 open roles, with posted comp from $200k–$255k and 100% open to remote. Below: what it's like to work there, how it pays, and how hiring works.

Open roles
81
Posted comp range
$200k–$255k
Remote-friendly
100%

Open roles at Cerebras

81 live roles — click any row for the full posting.

What Cerebras does

Cerebras builds the Wafer-Scale Engine, marketed as the world's largest AI processor, plus a hosted inference-and-training service aimed at customers seeking an alternative to NVIDIA GPUs. The entire architecture bets that a single dinner-plate-sized wafer beats racks of discrete chips for large-model workloads. The company IPO'd on Nasdaq as CBRS in mid-2026 after years as a marquee private AI-chip name, led throughout by CEO Andrew Feldman. The competitive frame is explicitly NVIDIA, with Groq, SambaNova and other accelerator startups nearby. Wafer-scale is a genuinely differentiated stack, which means the engineering problems — yield, packaging, thermals, and a bespoke software toolchain — are unlike anything in a conventional GPU pipeline. That novelty is the draw for hardware engineers and, as covered below, also the source of an unusually concentrated customer base.

What it is like to work at Cerebras

Wafer-scale logistics mean hardware, packaging, reliability and operations roles are largely on-site at the Sunnyvale and Toronto sites; software roles often allow hybrid arrangements. JobsByCulture describes a strong mentorship culture with collaborative engineering teams, and the numeric sentiment backs a mixed picture: Glassdoor shows a 91% recommend rating but a 3.7 work-life-balance score across a roughly 57-review sample. Employees rate engineering impact highly while leadership ratings run weaker, and the dominant review themes are IPO-timing pressure and awareness of the customer-concentration risk. This is a place where the technical mission is compelling and the collaboration real, but where the cadence tightened around the public offering and where everyone is conscious of the business's lopsided revenue base.

What Cerebras pays

Member of Technical Staff (MTS) is Cerebras' primary engineering title, and base salaries for MTS roles run roughly $164K to $196K, with a Hardware Engineer at the L5 level around $200K. That is mid-range hardware compensation — higher than Graphcore or Zoox at comparable levels, but below Astera Labs' principal-tier ceilings. The larger lever is now equity: as a freshly public company, post-IPO stock is the real upside, and candidates with ASIC or accelerator-architecture backgrounds command a premium given how niche the wafer-scale stack is. Weigh the cash bands as solid-but-not-exceptional and treat the equity, subject to public-market volatility, as the component that could move total comp meaningfully.

How hiring works at Cerebras

The official interviewing guide describes a three-step software process that opens with a 45-minute exploratory call aligned to background and role, then technical coding, then a full onsite loop covering system design and values. Hardware candidates separately report a 45-minute initial round to start the loop. The bar is high and specific: because the wafer-scale stack diverges so far from a standard GPU pipeline, candidates need genuine depth in accelerator architecture and system-level software rather than generic breadth. Expect the technical rounds to probe how you reason about memory, dataflow and parallelism at scale. Preparation should center on the fundamentals of large-model acceleration, not leetcode volume alone.

Growth & trajectory

Momentum has been steep. Cerebras took $1.1B in October 2025 for runway, then closed a $1B Series H on February 3, 2026 at roughly a $23B post-money valuation, before pricing its IPO and seeing shares open at $336 against a $185 offer price on May 14, 2026. The IPO range had implied up to about $48.8B fully diluted. For hardware engineers this is one of the strongest comp-upside trajectories in the cohort, and the public listing creates real equity liquidity. The catch is that the valuation rests on an unusually narrow revenue base, so the trajectory is best read as high-reward and high-variance rather than a smooth, diversified growth story.

Risks to know

The central risk is customer concentration, and it is openly disclosed: the prospectus names G42 as 87% of H1 2024 revenue and 91% of accounts receivable as of December 31, 2024. Analyst commentary has also flagged that CEO Andrew Feldman previously pleaded guilty to circumventing accounting controls at Riverstone Networks in the 2000s, and Pomerantz Law Firm has publicized a securities-fraud investigation on behalf of investors. None of this is hidden — it is in the filings and the press — which means candidates can underwrite it deliberately. If single-customer dependency or post-IPO governance scrutiny would keep you up at night, this is the profile to study most carefully before signing.

Who thrives at Cerebras (and who should not)

Thrives: an engineer energized by wafer-scale, customer-deployed inference and a collaborator-heavy, mentorship-driven culture, who is comfortable in a company whose IPO and concentration risks are spelled out in the S-1. Unlike pure-software AI, hardware roles here are physically tied to wafer bring-up and on-site labs, so lab proximity is part of the job. Should not apply: anyone whose career thesis depends on customer diversity, anyone allergic to a single-customer revenue model, or anyone who would be uncomfortable working somewhere with active securities scrutiny and the pressure cadence of a newly public company.

Roles Cerebras is hiring for

The roles Cerebras is most actively hiring right now in our index, with a live count and the salary guide for each:

The full board of open roles — with comp and location on every posting — is at the top of this page.

The signals behind this page

The hiring picture here is read from 81 live Cerebras postings in our index (refreshed weekly); 100% are remote-friendly, and in a recent sample 22 disclose a pay-transparency band. The culture, growth, and interview detail above is researched and cited; the open-roles board is live from our jobs index.

Sources

Prep for a Cerebras interview

Landed scores your readiness against real AI-native roles and drills the interview until you walk in ready.

Frequently asked

Is Cerebras a good place to work as an engineer?

Wafer-scale logistics mean hardware, packaging, reliability and operations roles are largely on-site at the Sunnyvale and Toronto sites; software roles often allow hybrid arrangements. JobsByCulture describes a strong mentorship culture wit

How many open roles does Cerebras have?

Our index tracks 81 live Cerebras roles right now, refreshed daily.

What does Cerebras pay?

Posted total comp spans $200k–$255k across levels for roles that disclose a band. See the per-role salary guides for percentiles.

Does Cerebras hire remote?

Yes — about 100% of Cerebras's current openings are remote-friendly.

Related