Machine Learning Engineer II

📋 Description

• • Design and implement systems that combine rules, models, feature engineering, and business/product inputs into an email detection product, with senior engineer guidance.
• • Understand features that distinguish safe emails from email attacks and how the model stack enables detection of these threats.
• • Identify and recommend new feature groups or machine learning model approaches to significantly improve detection efficacy for specific attack categories.
• • Work with infrastructure and systems engineers to productionize signals that feed into the detection system.
• • Write production-grade code with emphasis on testability, readability, edge case handling, and error resilience.
• • Train machine learning models on well-defined datasets to improve efficacy against specialized attack types.
• • Actively monitor and improve false negative (FN) rates and overall detection efficacy for message detection product categories through feature engineering, rule updates, and ML modeling.
• • Analyze false negative and false positive datasets to categorize capability gaps and recommend short-term feature and rule improvements to enhance detection performance.
• • Contribute to other areas of the stack, including building and debugging data pipelines and presenting results to customers through internal tools when needed.
• • Build and maintain automated model retraining pipelines encompassing data analytics, data generation, modeling, production evaluation, and automated deployment stages.
• • Model communication patterns to establish enterprise-wide baselines and incorporate these as robust signals in detection systems.
• • Combine message-level signals (e.g., presence of specific phrases), sender-level signals (e.g., frequency of sender), and recipient-level signals (e.g., likelihood of receiving safe messages) into precise detection systems.
• • Operate a detection decisioning system at extremely high recall while minimizing disruptions to customer workflows.
• • Understand the domain of false negatives — current and future attacks that could cause significant customer workflow disruption — and help define the technical roadmap to address them.
• • Apply a systematic approach to debug both data and system issues within machine learning and heuristic models.
• • Use SQL, pandas, and Spark frameworks to build data and metric generation pipelines and answer critical questions about system efficacy or counterfactual treatments.
• • Ensure all models and systems are reproducible with stable, production-level training and evaluation pipelines.