Machine Learning Engineer

This is a fully remote position, open to applicants in Maryland, +2 more states.

📋 Description

• Enhance ZoomInfo's data graph by incorporating information from lesser-known companies that have minimal public presence, including extracting details on leadership, locations, and products from their websites while identifying outdated records.

• Forecast headcount and revenue for less-documented companies utilizing techniques such as gradient-boosted trees, regression with missing inputs, and calibrated uncertainty assessments.

• Evaluate whether records pertain to the same company or individual and assess the outcomes of incorrectly merged or split records.

• Derive buying intent from multilingual and noisy web content to facilitate propensity scoring, lookalike retrieval, and contact suggestions.

• Create agents that conduct research on companies and cite their sources, while also designing evaluations that differentiate accurate results from those that simply completed processing.

• Simplify large models into more compact and cost-efficient versions, taking ownership of quantization and serving processes.

• Transform ambiguous, high-stakes challenges from undefined stages to shipped products.

• Elevate the engineering standards of the team through thorough review, design contributions, and mentorship.

⛳️ Requirements

• Proven experience in deploying machine learning systems into production and managing them post-launch.

• Demonstrated technical leadership as an individual contributor, involving direction setting, design review, and mentoring responsibilities.

• Expertise in classical machine learning beyond just language models, covering supervised learning and feature engineering on extensive and complex tabular datasets.

• Application of statistics in decision-making processes, including experiment design, statistical inference, and managing calibrated scores in scenarios of class imbalance.

• Experience in large-scale deployment of language processing, such as text classification, information extraction, and entity linking across substantial volumes of multilingual and noisy text.

• Background in building and managing LLM agents or multi-step systems in a production environment, which includes tool and context design, trace-based failure analysis, and evaluation utilizing LLM judges validated against human labels.

• Strong proficiency in production-level Python.

• Solid skills in SQL.

• Experience in distributed data processing.

• Regular use of AI coding tools while maintaining rigorous output reviews.

• Familiarity with ranking and retrieval concepts, including embeddings, learned re-ranking, recall@k, MRR, and nDCG (preferred).

• Experience in propensity modeling, clustering, or entity resolution with messy, real-world data (preferred).

• Experience in training and serving open-weight models using PyTorch or similar frameworks (preferred).

• Experience in defending LLM systems against adversarial inputs and prompt injection, or involvement in web-scale information extraction, knowledge graphs, or user memory for agents (preferred).

🏝️ Benefits

• Comprehensive benefits package.

• Holistic programs for mind, body, and lifestyle aimed at promoting overall well-being.

• Additional compensation may include bonuses, commissions, and equity options.