You'll set the technical direction for the intelligence layer every ZoomInfo AI agent reasons over — what's true about companies and the people in them, how they relate, and what they're buying. As a senior individual contributor with no direct reports, you will own strategy across the data graph, entity resolution, and buying intent, staying hands-on in the hardest of these problems. You'll decide when classical machine learning, statistics, or language models is the right tool, and turn good judgment into standards other teams adopt.

What You'll Do

• You will set technical strategy for the data graph, entity resolution, buying intent, and agent context, deciding which problems to take on, in what order, and how the pieces connect.
• You will frame and ship problems nobody has framed yet, such as extraction for companies with no public footprint or semantic intent at web scale, deciding what not to build as deliberately as what to build.
• You will define evaluation standards that other teams adopt, including sampling methodology, human labels, validated LLM judges, and release gates, and know where each one breaks.
• You will define the data contracts between the data graph, intent signals, and the agents that consume them, including agent memory, keeping the system cost-effective to run across the full dataset.
• You will turn good choices into defaults across the group: the baseline every model must beat, how uncertainty is reported, and which metrics get retired when they mislead.
• You will lead design reviews, mentor senior engineers, shape hiring, and represent the team's technical judgment to product and leadership.

What You Bring

Must-Have:

• You have set technical direction as an individual contributor for a problem area spanning more than one team, with the resulting machine learning systems in production and owned after launch — depth of impact matters more than years or past titles.
• You bring breadth across classical machine learning on large, messy tabular data, applied statistics, language processing at scale, and LLM agents or multi-step systems, with recognized depth in at least one.
• You have defined and driven adoption of evaluation standards beyond your own team for systems with no single right answer, including LLM judges validated against human labels and results reported with their limits.
• You have a track record of changing how a group builds or measures — for example, introducing a baseline or review practice that other teams now use.
• You have made cost and capacity decisions for model training and serving, including when to self-host, distill, or call a hosted model, with cost tracked per unit of work.
• You are hands-on in production Python and strong SQL with distributed data processing experience, and you use AI coding tools daily with rigorous review of their output.

Preferred:

• You have experience with ranking and retrieval, including embeddings, learned re-ranking, and metrics such as recall@k, MRR, and nDCG.
• You bring entity resolution, knowledge graph, or web-scale information extraction experience on messy, real-world data.
• You have trained and served open-weight models in PyTorch or an equivalent framework, including GPU capacity planning.
• You have built user memory or context systems for agents, or defended LLM systems against adversarial inputs and prompt injection.

#LI-Remote 

#LI-VC1