Machine Learning Engineer

Role Overview

Join Abnormal AI as a Machine Learning Engineer on the Message Detection – Attack Detection team. You will help build a high-recall Detection Engine that combines rules, ML models, feature engineering, and business inputs into an email detection product.

Responsibilities

• Design and implement systems that combine rules, models, feature engineering, and business inputs for email/message detection.
• Train and improve models on well-defined datasets, especially for specialized attacks.
• Actively monitor and improve detection performance, with emphasis on reducing false negatives and improving overall efficacy.
• Analyze datasets to enhance detection effectiveness.
• Contribute to other parts of the stack and support broader detection system improvements.

Requirements

• 3+ years experience designing, building, and deploying ML applications in areas such as:
  • text understanding, entity recognition, NLP
  • computer vision
  • recommendation systems
  • search
• Ability to understand business requirements and design a simplest yet generalizable ML model/system.
• Experience with big data, statistics, and machine learning.
• BS in Computer Science, Applied Sciences, Information Systems, or related engineering field.
• Strong data analytics experience, using SQL + pandas + Spark to build data and metric generation pipelines and answer questions about system efficacy/counterfactual treatments.
• Systematic debugging skills for data and ML/heuristics model issues.
• Strong software engineering habits: readable, structured, well-tested, efficient code.
• 1+ years writing stable, production-grade pipelines for training/evaluation to produce reproducible models and metrics.
• Fluency in Python and ML toolkits such as NumPy, scikit-learn, PyTorch, TensorFlow.
• Experience with algorithms and optimization.
• MS degree (Computer Science, Electrical Engineering, or related) is preferred/expected.

Nice to Have

• Not explicitly stated beyond the advanced degree signal and breadth of domain experience.