Чем предстоит заниматься
This is a high-ownership role at the intersection of agent systems, retrieval, and product. You will turn frontier-model capabilities into dependable, production-grade agent behavior, and you will own that behavior end to end, from the prompts and tools to the evaluation that proves it works
Design and build AI agents that plan, retrieve, and reason over real-world information to complete open-ended tasks
Build the tool interfaces and context engineering that let frontier models use our search and other tools effectively
Mine and analyze usage data to build agents that learn and improve continually from how they are used
Turn new model capabilities into reliable product features, and own them from prototype to production
Define the evaluations, metrics, and guardrails that prove an agent is accurate, grounded, and safe
Improve agent quality across reasoning, planning, tool use, and grounding against real user tasks
Build the backend and infrastructure that run agents reliably under high volume
Collaborate with the search, ML, and product teams to make agent and platform capabilities reinforce each other