Product Manager - AI Evaluations Tooling

Relativity ODA LLC; United States; Hybrid; USD 150,000 - 210,000; Full time

Job summary

Relativity is seeking a Product Manager to own the strategy and roadmap for the Evaluations Platform, enabling AI product teams to measure, understand, and improve AI performance. You will drive continuous discovery with scientists, engineers, and PMs, shape the agentic testing system, and ensure fast, safe deployment of capabilities at scale.

You will partner with legal and product stakeholders to define evaluation tasks, establish error criteria, and maintain dashboards that surface product

Qualifications

• 4+ years of product management at a technology company.
• 2+ years on technical, platform, infrastructure, or data/ML products.
• Strong understanding of the software development lifecycle and agile practices.
• Experience conducting user discovery with technical audiences and translating findings into roadmap decisions.
• Excellent written and verbal communication skills, with the ability to explain complex technical concepts to both technical and non-technical audiences.

Responsibilities

• Own the product vision, strategy, and roadmap for Relativity's Evaluations Platform across its pillars.
• Run continuous discovery with applied scientists, application engineering teams, and PMs to measure AI quality.
• Lead the agentic testing system and ship new capabilities safely and quickly.
• Define SME tasks, build authoring/approval workflows for experts.
• Own team-facing quality dashboards to show performance and changes since last release.
• Lead build-vs-buy evaluations for tracing infrastructure and eval tooling.
• Collaborate with engineering and applied science on judge calibration for gating deployments.
• Clearly articulate technical tradeoffs to stakeholders across levels.

Skills

LLM-based products

Platform/DevTools PM

Technical customers

Agile practices

User discovery

Clear communication

Job description

Posting Type

Remote/Hybrid

Job Overview

Relativity's AI portfolio spans 20+ products and we are moving quickly into agentic capabilities. As that portfolio grows, so does the opportunity to give every team building AI at Relativity a shared, standardized way to understand and improve their own AI performance. The Evaluations Tooling team exists to build that foundation: a platform that standardizes how we evaluate AI, accelerates how quickly teams can ship it with confidence, and turns AI quality into shared evidence that applied scientists, application teams, product managers, and legal experts can all act on.

We are looking for a Product Manager to own the product strategy and roadmap for Relativity’s evaluation platform. Your mission is to enable every team at Relativity building AI products to understand, measure, and improve their own AI performance. That means productizing what applied scientists have already validated into a platform that application teams, PMs, and legal subject‑matter experts can use directly; building a world‑class agentic testing system that lets us ship new agentic capabilities quickly and confidently; and using production monitoring and team‑facing dashboards to move the organization from reacting to quality problems to proactively getting ahead of them.

This is a platform PM role with unusually direct leverage. A successful candidate is energized by deeply technical customers, comfortable making build-vs-buy calls in a fast-moving vendor landscape, and able to hold a categorical bar ("every AI product ships behind elite evals") while sequencing pragmatically to get there.

Job Description and Requirements

Role Responsibilities

• Own the product vision, strategy, and roadmap for the Evaluations Platform across its pillars: offline evaluation pipeline, online evaluation and production monitoring, eval discovery and reuse, SME authoring and approval, and agentic evaluation.
• Serve every team building AI at Relativity. Run continuous discovery with applied scientists, application engineering teams, and product PMs to understand how each of them experiences AI quality today and design the platform so teams can measure and improve their own performance without routing every question through an applied scientist.
• Lead the orchestration of a world‑class agentic testing system: trajectory evaluation, tool-call correctness, intermediate‑state rubrics, long‑running judge orchestration, and simulation of multi‑step flows. Make it fast and safe to ship new agentic capabilities and make the platform a competitive advantage in how quickly Relativity can iterate on agents.
• Work closely with legal subject‑matter experts to define the tasks our AI is evaluated against, what a correct outcome looks like, where the hard cases are, and how judgment criteria should be expressed so they can be encoded, scored, and reused across products. Build the SME authoring and approval workflow around how experts work.
• Own the team‑facing quality dashboards. Every AI product team should be able to open a view and understand their own performance, where they stand against their quality bar, and what changed since last release, with no engineering help required.
• Lead build‑vs‑buy evaluations for tracing infrastructure and commercial eval tooling and run a recurring review as the vendor landscape shifts.
• Partner with engineering and applied science on judge calibration so LLM‑as‑judge is trustworthy enough to gate deployments.
• Clearly articulate technical tradeoffs to stakeholders ranging from applied scientists to attorneys to executive leadership.

Preferred qualifications:

• Experience with LLM‑based products and a working understanding of how they are evaluated (rubrics, LLM‑as‑judge, offline vs. online evaluation). Direct experience with agentic systems is a plus, not a requirement.
• Experience as a platform, developer‑tools, or data/ML PM serving technical customers, or a demonstrated ability to learn a technical domain quickly and earn credibility with engineers and scientists.
• Comfort bringing non‑engineering domain experts into technical workflows and translating their judgment into something a system can act on.
• Strong analytical instincts: sets metrics, tests assumptions, and changes course when the evidence says to.

Mi nimum qualifications:

• 4+ years of product management experience at a technology company, with at least 2 years on technical, platform, infrastructure, or data/ML products.
• Solid understanding of the software development lifecycle and agile practices.
• Experience conducting user discovery with technical audiences and translating findings into roadmap decisions.
• Excellent written and verbal communication skills, with the ability to explain complex technical concepts to both technical and non-technical audiences.

Relativity is committed to competitive, fair, and equitable compensation practices.

This position is eligible for total compensation which includes a competitive base salary, an a