Lead Scientist: Efficient LLM Inference & Optimization

Nebius Group; Greater London; Hybrid; GBP 90,000 - 150,000; Full time

Benefits offered by this job

Competitive compensation

Job summary

Nebius Group, a cloud infrastructure leader, seeks a PhD-level researcher to drive efficient LLM/VLM inference research and translate findings into production-ready solutions. You will lead experiments, build prototypes with PyTorch/Triton, and collaborate with engineers to optimize throughput, latency, and cost across distributed systems.

You will publish results, mentor peers, and contribute to open-source artifacts while shaping research priorities with measurable production impact.

Qualifications

• PhD in computer science, ML systems, or closely related field.
• Strong publication record or equivalent artifacts in ML and efficient inference.
• Proficient in Python and PyTorch with ability to implement experiments and prototypes.
• Deep knowledge of LLMs, VLMs, transformer inference, quantization, and production-serving trade-offs.
• Strong experimental design skills including ablations, baselines, metrics, and failure analysis.
• Excellent written and verbal communication.

Responsibilities

• Lead research projects in efficient LLM/VLM inference from hypotheses to production handoff.
• Develop and evaluate methods: quantization, QAT, distillation, KV-cache optimization, and model/runtime co-optimization.
• Build prototypes using PyTorch/Triton/CUDA tooling and collaborate with engineers to productionize.
• Study routing, scheduling, and cache strategies to optimize latency and cost.
• Evaluate multi-node inference, WideEP, and expert placement for performance and efficiency.
• Share results via internal reports, blogs, papers, and open-source artifacts; mentor teammates.

Skills

Python

PyTorch

Experiment design

LLM inference

Research

Education

PhD in CS / ML systems

Tools

PyTorch

Triton

CUDA

TensorRT

Job description

Nebius Group, a cloud infrastructure leader, seeks a PhD-level researcher to drive efficient LLM/VLM inference research and translate findings into production-ready solutions. You will lead experiments, build prototypes with PyTorch/Triton, and collaborate with engineers to optimize throughput, latency, and cost across distributed systems.

You will publish results, mentor peers, and contribute to open-source artifacts while shaping research priorities with measurable production impact.