Senior Applied Scientist, Efficient LLM Inference & Model Optimization

nebius; United States; Remote; USD 180,000 - 260,000; Full time

Job summary

nebius seeks a Senior Applied Scientist to advance efficient LLM and VLM inference within our AI cloud platform. The role blends deep research with production engineering impact, translating bottlenecks into scoped studies and shipping results into deployed inference capabilities.

Collaborate with ML engineers to turn research prototypes into production components, define research programs, and publish credible work when appropriate. A PhD and strong Python/PyTorch skills are required.

Qualifications

• PhD in CS/ML systems or related field.
• Strong publication record or equivalent artifacts.
• Strong coding in Python and PyTorch.
• Deep understanding of LLMs, VLMs, transformer inference, decoding algorithms, model compression, quantization, and production-serving tradeoffs.
• Strong experimental design skills, including ablations, baselines, metrics, statistical reasoning, and failure analysis.

Responsibilities

• Own focused research projects from hypothesis through experiment, ablation, prototype, and production handoff.
• Prepare internal reports, technical blogs, or external papers when work is publication quality.
• Partner directly with machine learning engineers to turn research prototypes into usable production components.
• Define and execute research programs in efficient LLM and VLM inference with measurable production impact.
• Invent, evaluate, and productionize methods for quantization, QAT, distillation, speculative decoding, KV-cache reuse and compression, long-context inference, MoE routing, and model/runtime co-optimization.
• Build prototypes in PyTorch, Triton, CUDA-adjacent tooling, or inference-serving frameworks, and collaborate on productionization.
• Design rigorous evaluations covering quality, latency, throughput, numerical stability, memory footprint, tail latency, and cost per token.
• Mentor engineers and scientists on experimental design and model/system tradeoffs.

Skills

Python

PyTorch

Research

Experimental design

Communication

ML systems

Education

PhD in CS / ML systems or related field

Tools

PyTorch

Triton

CUDA tooling

Inference serving frameworks

Job description

Role overview

A Senior Applied Scientist role focused on efficient LLM and VLM inference and model optimization within an AI cloud platform. The position blends rigorous research with production engineering impact: turning frontier inference bottlenecks into well-scoped research problems, publishing credible work, and shipping the results into deployed inference capabilities. It is explicitly not a papers-only role, and close collaboration with machine learning engineers is expected throughout.

Responsibilities

• Own focused research projects from hypothesis through experiment, ablation, prototype, and production handoff.
• Prepare internal reports, technical blogs, or external papers when work is at publication quality.
• Partner directly with machine learning engineers to turn research prototypes into usable production components.
• Define and execute research programs in efficient LLM and VLM inference with measurable production impact.
• Invent, evaluate, and productionize methods for quantization, QAT, distillation, speculative decoding, KV-cache reuse and compression, long-context inference, MoE routing, and model/runtime co-optimization.
• Build prototypes in PyTorch, Triton, CUDA-adjacent tooling, or inference-serving frameworks, and collaborate on productionization.
• Design rigorous evaluations covering quality, latency, throughput, numerical stability, memory footprint, tail latency, and cost per token.
• Mentor engineers and scientists on experimental design and model/system tradeoffs.

Requirements

• PhD in computer science, machine learning, ML systems, computer systems, computer architecture, electrical engineering, applied math, or a closely related field.
• Strong publication record or equivalent research artifacts in ML, ML systems, efficient inference, model compression, quantization, distillation, or serving systems.
• Strong hands-on coding ability in Python and PyTorch, with the ability to move from idea to experiment to prototype quickly.
• Deep understanding of LLMs, VLMs, transformer inference, decoding algorithms, model compression, quantization, and production-serving tradeoffs.
• Strong experimental design skills, including ablations, baselines, metrics, statistical reasoning, and failure analysis.
• Excellent written and verbal communication.

Nice to have

• First-author publications at NeurIPS, ICML, ICLR, MLSys, ACL, EMNLP, ASPLOS, OSDI, SOSP, ISCA, HPCA, or comparable venues.
• Experience deploying ML models or inference optimizations in production.
• Experience with vLLM, SGLang, TensorRT-LLM, NVIDIA Dynamo, FlashAttention, FlashInfer, Triton, CUDA, or PyTorch internals.
• Experience with post-training, SFT, DPO, RLHF, RLAIF, preference optimization, or synthetic data generation tied to inference quality or efficiency.
• Open-source research artifacts, widely used benchmarks, technical blogs, or invited talks in efficient AI systems.

Benefits and work setup

• Competitive compensation.
• Career growth and learning opportunities in a fast-moving AI environment.
• Flexibility and meaningful ownership over research direction.
• Collaborative and innovative culture across international teams.
• Opportunity to work on high-impact AI infrastructure projects.