6 hours ago
Base Salary
$128k - $201k/yr
Responsibilities
- Extend the ZoomInfo data graph by extracting leadership, locations, and products from company websites and detecting stale records.
- Estimate headcount and revenue for under-documented companies using gradient-boosted trees, regression with missing inputs, and calibrated uncertainty.
- Build entity-resolution systems that determine whether records describe the same company or person while measuring false merges and false splits.
- Infer buying intent from multilingual web content for propensity scoring, lookalike retrieval, and contact recommendations.
- Build production research agents with source citations and design evaluations using LLM judges validated against human labels.
- Distill large models into smaller cost-effective models and own quantization and serving.
- Take ambiguous, high-stakes problems from definition through production and provide technical leadership through design reviews and mentorship.
Requirements
- Experience taking machine learning systems to production and owning them after launch, with individual-contributor technical leadership including setting direction, design review, and mentoring.
- Classical machine learning expertise beyond language models, including supervised learning and feature engineering on large, messy tabular data.
- Experience applying statistics to experiment design, statistical inference, and calibrated scores under class imbalance.
- Experience deploying language processing at scale, including text classification, information extraction, and entity linking over multilingual, noisy text.
- Experience building and operating production LLM agents or multi-step systems, including tool and context design, trace-based failure analysis, and evaluation for systems without a single correct answer.
- Proficiency in production Python, strong SQL, and distributed data processing experience.
- Daily use of AI coding tools with rigorous review of their output.
- Preferred: experience with ranking and retrieval, embeddings, learned re-ranking, recall@k, MRR, and nDCG.
- Preferred: propensity modeling, clustering, entity resolution, web-scale information extraction, knowledge graphs, or user memory for agents.
- Preferred: training and serving open-weight models in PyTorch or an equivalent framework, with attention to cost per unit of work.
- Preferred: defending LLM systems against adversarial inputs and prompt injection.
Benefits
- Remote work is indicated by the #LI-Remote posting marker.
- Comprehensive benefits and holistic mind, body, and lifestyle programs are offered.
- Additional compensation may include bonus, commission, equity, and other benefits.
Categories
About ZoomInfo
ZoomInfo builds a subscription go-to-market platform for B2B sales and marketing teams, combining company and contact data, intent signals, and sales engagement tools. Its products help organizations find prospects, align GTM motions, and automate outreach, and are used by more than 35,000 companies. The public company (NASDAQ: GTM; also traded in Frankfurt) is headquartered in Vancouver, Washington, and grew out of DiscoverOrg’s acquisition of ZoomInfo.