Machine Learning Architect - IV

Job Title

Hybrid: 3 to 5 days a week in the office (Basking Ridge, NJ or Rochelle Park, NJ)

Experience

• 7+ years experience in designing and developing enterprise class AI platforms and solutions
• 3+ years of experience with enterprise fully automated model and risk management solution.
• 3+ years implementing data ops, ML ops
• MS or BS in Computer Science, Information Science, Engineering or other related field

Skills

• Deep understanding and hands on experience with ML engineering techniques and tools including hands on experience with ML operations.
• Experience with the primary managed data services within Google Cloud Platform, including AI Vertex, Cloud Bigtable, Cloud Spanner, Cloud SQL, or BigQuery
• Proficient in data science workbenches such as Domino, container platform such as K8s/Docker, Core Java, J2EE, JSP, Servlet, Node.js, Angular
• Proficient in big data technologies, data transport (Pulsar/Kafka), Spark, Jupyter/Python
• Experience working with multiple databases: Cassandra, PostGreS, Teradata. and NoSQL and RDBMS technologies container platform such as K8s/Docker
• Experience with various agile methodologies and tools: JIRA, Confluence, Gitlab, CICD, etc.
• Exposure to product based development methodology is desirable
• Strong leadership, communication, persuasion and teamwork skills

ML Model Management Platform Strategy

• Define and architect comprehensive model management framework across these 4 major areas
• Monitor data quality - monitor drift in data quality.
• Monitor model quality - monitor drift in model quality metrics, such as accuracy.
• Monitor bias drift for models in production - monitor bias in model's predictions.
• Monitor feature attribution drift for models in production - monitor drift in feature attribution.

Technology / Execution

• Build and implement a platform for seamless integration and interface with existing batch and realtime ML systems to enable track performance metrics and verify the accuracy of predictions
• Design/implement a clean UI so that data drift, model quality, and other health statistics are provided in an easy-to-understand interface to enable quick assessment of the business impact and initiate proactive actions
• Implement appropriate notifications, alerts for both upstream and downstream systems