Staff / Principal Applied AI Researcher (Agentic Search)

About Nebius

Nebius is leading a new era in cloud infrastructure for the global AI economy. We are building a full-stack AI cloud platform that supports developers and enterprises from data and model training through to production deployment, without the cost and complexity of building large in-house AI/ML infrastructure.

Built by engineers, for engineers. From large-scale GPU orchestration to inference optimization, we own the hard problems across compute, storage, networking and applied AI.

Role Overview

We are seeking a Staff or Principal Applied AI Researcher to join a fast growing team building an agent native search platform - the web access layer for AI systems.

You can think of this as Google for AI agents: a system designed for machines, not humans. We are building agentic search, where AI systems actively plan, retrieve, evaluate, and refine information rather than simply returning results. As AI becomes the primary interface to the web, this layer will replace the role of traditional search engines.

We are designing how AI agents - not humans - retrieve, evaluate, and reason over web data in real time, under strict latency and reliability constraints. This means solving retrieval and ranking under entirely new access patterns and at significant scale, with systems operating over constantly changing, unstructured data and serving tens of thousands of production workloads 24 by 7.

This role comes with ownership over key parts of our applied AI research direction and system design, with a strong expectation of defining new approaches and shipping measurable impact in production.

What you'll work on

• Designing agent native retrieval systems optimised for machine consumption rather than human search UX
• Building systems where LLMs iteratively plan, query, refine, and reason over results
• Developing ranking and retrieval approaches for multi step, agent driven workflows under real world constraints

Your responsibilities

• Drive applied research and technical direction across retrieval and ranking systems
• Design and evolve multi stage retrieval architectures (query understanding, rewriting, reranking, iterative retrieval)
• Develop methods for grounding LLMs in real time web data at scale
• Define and implement new evaluation paradigms and metrics for agentic systems, where correctness is not reducible to clicks
• Lead experimentation on modern retrieval approaches (embeddings, hybrid search, reranking) and bring them into production
• Analyse trade-offs across relevance, latency, and cost at scale
• Work closely with engineering to deploy systems in high throughput, low latency environments