Applied Researcher – AI Expert

Applied Researcher – AI Expert

Location: Hybrid | Bellevue, WA (downtown)

Our client is seeking an experienced AI Expert / Applied Researcher to help shape how a fast-growing technology organization understands and applies the rapidly evolving landscape of AI models, architectures, inference technologies, and accelerator systems.

This role sits at the intersection of AI research, systems engineering, and infrastructure strategy. You will evaluate emerging technologies, translate research into practical engineering implications, and help guide decisions around AI infrastructure, inference optimization, model serving, accelerator platforms, and the economics of delivering AI workloads at scale.

This is a highly strategic individual contributor role with broad technical influence. You will work closely with engineering, product, infrastructure, finance, and commercial teams to help determine what technologies to build, adopt, partner for, or invest in.

• Hold the company's view of where data center infrastructure is going.

• Track vendor and hyperscaler roadmaps, research, standards work, and the startup and venture landscape across power and energy, cooling and thermal, construction and delivery, rack and hall architecture, siting and regulation, and the economics that connect them. Right now that means questions like whether facility-level 800 VDC becomes the standard, how far liquid cooling has to go as racks move from roughly 200 kW toward a megawatt, how much of a build program can be moved into a factory, and whether behind-the-meter generation beats waiting in the interconnect queue.
  
  Those specific questions will have changed within a year — holding the current version of them is the job.

• Formulate and validate the product and engineering thesis. Turn that view into a defensible position on what we build, buy, or partner for, pressure-tested against cost, schedule, and what the physical plant can actually support — and say so plainly when the evidence does not hold up.

• Own the technical reference view of a client's hall at each GPU generation: density, power envelope, cooling topology, and what our operating and prospective sites can and cannot absorb. This includes getting on site to see them.

• Influence the numbers finance, pricing, and sales depend on: cost per rack, per MW, and per GPU-hour, and how they move with density, cooling approach, and silicon mix. Be able to defend them under challenge.

• Make the work land commercially. Support sales and delivery in technically demanding customer and partner conversations, feed product and go-to-market with what we can credibly offer and when, and provide technical diligence on infrastructure partners, colocation providers, vendor reference designs, and prospective tuck-in targets.

• Ideally also validate the power picture, since it is the constraint that governs everything else — interconnect availability and queue position, utility and PPA structures, on-site generation, long-lead equipment, and how all of it sets site selection and build sequencing.

• Significant hands-on experience with modern AI models,

• Depth in inference rather than training alone,

• Working fluency in one or more accelerator ecosystem, and

• Hands-on depth at the compiler, kernel, or runtime layer (CUDA, Triton, ROCm/HIP, XLA, or similar).

• Working fluency in more than one silicon ecosystem. CUDA plus ROCm, Cerebras, or another accelerator experience combined with a realistic view of what portability actually costs.

• Fluency with the landscape you would be scanning: the frontier labs and open-weight model providers, the serving and inference startups, the silicon vendors, and the…