How to get Fractile jobs after its $720M raise

What did Fractile announce, and why does it mean more jobs?

Fractile is an AI chip startup based in London and Bristol. On 11 November 2026 it announced a $720 million Series C, and the money is meant to get its first chips and systems shipped next year. That means hiring. Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint led the round, and the UK government's Sovereign AI fund and the British Business Bank also invested. Reported valuations range from about $6.5 billion to $7 billion. That's a steep climb for a company founded in 2022 that only raised a $220 million Series B in May. Very few British hardware startups have ever had this much money to spend, and hardware money mostly goes on engineers.

The product explains who gets hired. Fractile builds chips for AI inference, which means running trained models rather than training them. Its pitch is that conventional hardware wastes time and energy moving data back and forth between memory and compute. Its design stacks DRAM directly on the chip and pairs that silicon with Fractile's own machine learning compiler. The company says this can sharply cut the cost of running large language models. Turning that design into working systems in data centres by 2027 takes a lot of very specific people: architects, RTL designers, verification engineers, physical designers, test engineers, compiler writers, runtime and driver developers, and the infrastructure engineers who keep the build pipeline running.

There's a catch for job seekers. A big funding announcement brings a flood of applications, and a startup that just made headlines will get CVs from everyone who's ever touched a GPU. The Fractile careers page was already listing around 25 open roles across Bristol, London and Taipei before this round. Hiring is already under way, so you're joining a queue that has already formed. Here's my blunt advice: don't take the news as a reason to apply for everything. Pick one or two roles where your experience genuinely overlaps, tailor your CV hard, and apply before the backlog grows. At a company this specialised, one focused application to the right ad beats five generic ones every time.

Which Fractile jobs are likely to open first?

Most new Fractile jobs will probably fall into four families: silicon design and verification, ML compiler and runtime software, systems software such as drivers and infrastructure, and semiconductor test and production. That's not guesswork. It matches roles the company has already advertised, including Senior ML Compiler Engineer, Senior ML Runtime Engineer, Senior Linux Kernel Driver Engineer, Senior Rust Software Engineer, Lead Physical Design Engineer and Semiconductor Test Engineering Team Leader. Fractile's first silicon and systems are due next year, and when a chip moves from design to deployment, hiring usually shifts towards bring-up, test, software maturity and customer-facing engineering. Listings for a developer experience engineer and a trade compliance manager suggest the company is already planning for customers and exports, not just tape-out.

There's also a useful signal for less senior candidates. Fractile has posted early-career architect roles (system, SoC and compute architect positions in London or Bristol), which is unusual for a deep-tech hardware company at this stage. Plenty of strong graduates assume AI chip startup jobs in the UK go only to people with a PhD or fifteen years at a big chip firm. That's not the full picture. If your master's or PhD project was hardware-focused, if you built a performance model of an accelerator, or if you wrote a cycle-accurate simulator, these roles deserve a serious look. Lead with the project, the modelling tools you used and what your results showed. Leave the long list of coursework out.

Location matters too. Most engineering roles are listed for London or Bristol, often with a choice between the two, and at least some are hybrid. The kernel driver role, for example, asked for three days in the office and two at home. Bristol makes sense given the city's long history in chip design. The Taipei office suggests close ties to Taiwan's semiconductor supply chain. If you need visa sponsorship, check each listing and ask early instead of assuming. Recent ads list equity, private medical cover, a contributory pension and 25 days' holiday plus bank holidays. After a round like this, the equity is worth more than it was, which makes these some of the more interesting London AI startup jobs to weigh up.

What should a chip design engineer CV show for Fractile?

A chip design engineer CV for Fractile should show that you've taken real silicon, or something close to it, through a named stage of the design flow, with the tools and the outcome spelled out. Hardware hiring managers read for scope first: which block, which process node, which stage, how big the team was. 'Worked on SoC design' tells them nothing. 'Owned RTL for a DMA engine in SystemVerilog, from microarchitecture spec to sign-off, on a 5nm tape-out' tells them almost everything they need. If you've been through tape-outs, say how many and what your part was. If you haven't, say exactly how far your block got and what you verified it against. A strong AI hardware engineer resume is specific from the first bullet.

Verification engineers need the same precision, plus metrics. Name your methodology and languages, such as UVM, SystemVerilog assertions and any formal tools you've used. Then put numbers on it: the coverage targets you closed, the bugs you caught before silicon, how much you cut regression runtime. For physical design, the story is timing closure, power and congestion, so give the frequency targets you met and the tools you ran. Test engineers should use the same vocabulary as Fractile's own test leadership listing: ATE programme development on platforms such as Advantest 93K or Teradyne, scan, ATPG and memory BIST. If those exact terms are in the ad and in your experience, they belong in your bullets, not buried in a skills list at the bottom.

Then add the part most hardware CVs leave out: why your work matters for AI inference. Fractile's whole thesis is about memory bandwidth and data movement. Put any experience with memory subsystems, DRAM interfaces, on-chip interconnect, dataflow accelerators or power-performance trade-offs near the top. Here's my contrarian take: don't pad your CV with generic AI buzzwords to look relevant. A silicon engineer who writes 'passionate about AI' in a summary sounds like someone who hasn't done AI hardware. One concrete bullet about cutting memory stalls on a matrix-multiply block does far more. If your older CV needs restructuring around those stages and results, the HRLens CV builder can rebuild it as a clean, ATS-friendly document through chat.

How should ML infrastructure engineers and embedded developers reframe their CVs?

ML infrastructure engineers should reframe their CVs around performance on hardware, not model accuracy or product features. Fractile's compiler listings ask for proven experience building ML compilers with MLIR or TVM, excellent C/C++, a good grasp of computer architecture and performance optimisation, and current knowledge of ML techniques. The ad also says the architecture is still being shaped, including the IR, the lowering strategy, the optimisation passes and the execution model. So the team wants people who have designed those pieces, not just maintained them. If you've written compiler passes, lowered operations to a new backend or profiled kernels on custom hardware, put that in the first two bullets of your latest role, with a speed-up or latency figure attached.

Coming from GPU work is fine as long as you translate it. If you've spent three years tuning CUDA kernels, writing Triton or optimising inference serving, you've built good instincts about memory hierarchy, batching and throughput. The framing doesn't carry over by itself, though. Swap 'improved model serving' for something like 'cut p95 inference latency 38% by fusing attention kernels and restructuring the KV-cache layout'. The numbers have to be your own, but that's the shape an accelerator team looks for. The same goes for ML runtime candidates: scheduling, memory allocation, host-device communication and multi-device orchestration are the keywords, and the closer your work sits to the hardware, the better your odds. We cover the same pattern in our guide to tailoring an ML infrastructure CV for a freshly funded AI startup.

Embedded and systems developers have a clearer path in than they might think. A chip is useless without drivers, firmware and tooling, and Fractile has advertised for a Senior Linux Kernel Driver Engineer and a Senior Rust Software Engineer. Your CV should name the subsystems you've worked on, such as PCIe, DMA, interrupt handling, memory management and device bring-up, plus the kernels and SoCs you shipped on. Upstream kernel contributions are gold, so point to the specific patch series instead of your whole profile. If you've written Rust in production, say where, why it replaced C, and what changed for reliability. Bring-up on unreleased silicon is especially valuable right now, because a company shipping its first chips next year needs exactly that.

How do you check your CV against a Fractile job description before applying?

The quickest check is to put the exact Fractile job ad next to your CV and confirm that every must-have tool and responsibility appears in your experience, in the same words, with evidence. Copy the text of the listing rather than the link, because sites like LinkedIn and Indeed block automated fetching. Highlight each hard requirement, whether that's MLIR, TVM, C/C++, the Linux kernel, UVM, scan or ATPG. Then go through your CV line by line. A requirement you meet but don't mention is an easy fix. A requirement you don't meet is a decision: if it's a deal-breaker, choose a different role; if it isn't, deal with it honestly, for example by pointing to a closely related tool you've used.

Formatting is the second filter, and it's where strong engineers lose out for silly reasons. Fractile's listings have appeared on standard hosted job boards, which turn your file into structured fields before a person reads it. Two-column templates, text boxes, icons and skill bars can scramble that process, so your carefully listed toolchain comes out garbled or disappears. Keep to one column and standard headings such as Experience, Skills and Education. Send a PDF or Word file with real selectable text, and write dates the same way throughout. Hardware CVs also tend to run long. Two pages is enough for most candidates, with publications and patents in their own short section rather than scattered through your roles.

Finally, run an objective check before you hit submit. With HRLens you can check whether your CV matches a specific Fractile job description. Upload your CV and paste the ad. You'll get a score out of 100, five category scores including ATS compatibility, and a list of skills the job asks for that your CV doesn't show. Every analysis also produces ATS-friendly rewritten versions in six templates that use the job's keywords, so you don't have to rebuild from scratch for each role. If you'd like a baseline first, run a free CV analysis and fix the obvious gaps. Fractile hiring in 2026 will move fast, so get your CV ready before the ad you want closes.

Frequently asked questions