Principal Applied AI Engineer - Entity Agents
Directly builds and ships LLM-based agent systems and evaluation harnesses; works on agent frameworks and deploying agents to an internal platform.
About the Role
Build and own the production machinery for ZoomInfo’s entity agents: complete the entity-agent pipeline, implement evidence adapters, run automated evaluation and gating, and deliver platform-ready agent deployments so researchers and PMs can operate agents without engineering support.
Job Description
Role
Senior Applied AI Engineer on Core Data responsible for productionizing agent-driven, AI-native data management for company and contact entities. The role focuses on completing the entity-agent pipeline, building evidence adapters, operating evaluation and gating systems, and deploying agents onto ZoomInfo’s Agentic Platform.
Key Responsibilities
- Run and schedule the benchmark and golden-gate evaluation system to automatically gate agent releases, track variance and drift, and enforce thresholds set by product.
- Complete pipeline stages (enricher fan-out, location sets, hierarchy batch files) and migrate classifiers to a shared production runner; measure throughput and unit cost per stage.
- Build evidence adapters (registries, live web, mail-tenant signals, external vendors) behind a single envelope with a replayable vendor store, per-vendor kill switches, and vendor-swappable design.
- Deploy company and contact agents onto the ZoomInfo Agentic Platform (ZAP) with ingest adapters, rollback paths, and coordinated platform/DOH upgrades.
- Build self-service tooling for researchers and product managers: review queues, tagging tools, blast-radius viewer, and batch runners; measure review yield and golden-record production.
- Own health and deployment of classifiers, agents hub, and judge platform; instrument cost, latency, and quality metrics and alerting.
Requirements
- 5+ years software engineering experience, including at least 2 years shipping LLM-based systems to production (agent loops, tool use, retrieval, structured output, and evaluation harnesses).
- Strong Python and TypeScript skills; comfortable across API layers, batch runners, and small internal web UIs.
- Experience building platforms for others to operate: adapters, runners, queues, and internal tools designed for replayability, idempotency, and observability.
- Working knowledge of data pipelines and entity resolution (ingestion, matching, serving) and operational trade-offs (false merges vs. missed matches).
- Proven track record of taking prototypes to production while preserving what prototypes proved; treat test sets and benchmarks as contracts.
- Instrumentation-first mindset: understand cost and latency budgets for shipped systems.
- Bias for action and ability to drive decisions across a distributed organization.
Preferred
- Experience with B2B company/contact data, identity/registry data, or entity resolution at scale.
- Experience with Claude or comparable frontier models, agent frameworks, and evaluation harnesses.
- Experience deploying agents onto internal agent platforms or orchestration layers.
- Background at AI-native startups, applied AI teams, or data infrastructure companies.
Team & Environment
- Reports to Senior Manager of Product for Core Data and partners closely with the Principal AI Engineer, product managers (evaluation & policy), company/person engineering teams, research graders/rulers, and the agentic platform team.
- Focus on running large-scale judge/eval workloads cost-effectively and making vendor integrations replaceable without extensive rewrites.