Principal Scientist, Agentic Systems Applied Science
Role details
Job location
Tech stack
Job description
We are seeking a Principal Scientist to establish how agentic AI systems are built, evaluated, and scaled at Wayfair. This role sits within the GST + SCRT Applied Research organization and is focused on one of the most consequential open problems in applied AI: how do you build autonomous systems that can reason, plan, and act across complex supplier and supply chain workflows - reliably, safely, and at scale?
This is paradigm creation work. You will not be inheriting a playbook; you will be writing it. As Wayfair's technical authority on agentic systems, you will define the architectural patterns, evaluation frameworks, and safety standards that downstream Applied Science and Engineering teams build on. Your work compounds across every autonomous system Wayfair builds next.
You'll act as the connective tissue between research, engineering, and product - translating foundational science into reusable patterns that accelerate teams across catalog, supply chain, and customer technology without requiring each to reinvent the same hard problems. This is a high-leverage role for someone who thrives at the intersection of deep technical rigor and real production stakes.
What You'll Do
- Research and prototype agentic architectures - including planning, memory, tool use, and multi-step reasoning - relevant to catalog and supply chain workflows.
- Define Wayfair's evaluation frameworks for agentic systems, moving beyond offline accuracy and A/B tests to assess reliability, safety, and business impact in non-deterministic environments.
- Establish company-wide standards for autonomy boundaries, failure modes, rollback procedures, and human-in-the-loop design - and drive adoption across technology teams.
- Build reusable agentic patterns that Applied Science and Engineering teams can adopt safely, reducing organizational debt and preventing fragmented, one-off implementations.
- Identify where agentic approaches outperform deterministic pipelines - and where they do not - serving as a system-level gate on investment decisions.
- Partner with downstream teams in Supply Chain, Search, and Customer Technology to pilot validated patterns and demonstrate measurable business impact.
- Influence platform strategy by providing input on tooling, evaluation infrastructure, and resource allocation decisions for agent development across the organization.
- Define a proactive research agenda that keeps Wayfair at the frontier of agentic AI capabilities as the field evolves.
Requirements
- Deep hands-on expertise in LLM-based systems, agent architectures, and multi-step reasoning pipelines - you have built these in production, not just studied them.
- Experience designing evaluation frameworks for complex, non-deterministic AI systems, with a strong understanding of where standard metrics fall short.
- A track record of technical influence across engineering, applied science, and product organizations without requiring direct organizational authority.
- The judgment to identify when agentic approaches should not be used - and the credibility to make that call stick with senior stakeholders.
- Comfort operating in ambiguous, greenfield problem spaces where the right framework does not yet exist.
- PhD or equivalent research experience in machine learning, AI, or a related field, combined with meaningful industry experience building systems at scale.
- Strong communicator with a proven ability to translate complex research findings for technical and non-technical audiences alike.