Inference Engineer (AI)

TL;DR

Inference Engineer (AI): Optimizing and scaling foundation model inference on Blackwell clusters with an accent on cost-per-token, throughput, and latency. Focus on solving complex systems challenges like disaggregated prefill/decode, KV-cache hierarchy, and low-precision MoE serving.

Location: Must be based in the Netherlands or Switzerland, with an expectation of spending at least 50% of time in the office.

Company

hirify.global is building a next-generation agentic clinical AI assistant designed to support clinicians with longitudinal patient context and complex diagnostic workflows.

What you will do

• Instrument and analyze the inference stack on Blackwell to optimize token cost, throughput, and latency.
• Tune scheduling and admission control to maintain cost efficiency across ramp-up and steady-state regimes.
• Manage the KV-cache hierarchy and optimize the prefill/decode split.
• Drive low-precision MoE serving while implementing quality regression gates.
• Collaborate with product and research teams to deploy new models and workloads.

Requirements

• Must be based in the Netherlands or Switzerland.
• Deep GPU systems experience, including kernel-level CUDA or Triton development.
• Proficiency with CUTLASS, FlashInfer/Flash Attention, and Nsight profiling.
• Proven experience shipping production inference stacks at scale (e.g., vLLM, SGLang, TensorRT-LLM).
• Strong understanding of roofline models, arithmetic intensity, and KV-cache costs.

Nice to have

• Experience with quantization kernels (FP8/FP4, AWQ/GPTQ).
• Expertise in MoE serving, including expert parallelism and routing.
• Experience scheduling shared training and inference workloads.
• Background in healthcare or regulated-deployment environments.

Culture & Benefits

• Competitive salary, pension plan, and 25 days of vacation.
• EUR 1000 annual learning and development budget.
• Regular offsites and team events.
• Annual commuting subsidy.
• Flexible work environment focused on autonomy and ownership.

Hiring process

• Screening call to align on motivation and professional goals.
• Technical take-home assessment.
• Technical assessment debrief to discuss problem-solving and team fit.
• Final onsite interview to discuss long-term alignment and impact.