Principal AI Data Platform Architect

Saviynt; El Segundo (CA); On-site; USD 240,000 - 304,000; Full time

Job summary

Saviynt seeks a Principal SWE to steer the AI Platform's data architecture, ensuring secure, scalable data flows from raw inputs to model signals. You will define standards for ML data pipelines, governance, and PII safeguards while maintaining cross-team collaboration across a large SaaS platform.

You will own Spark-based processing, streaming with Dataflow, and orchestration with Flyte and Kubeflow, plus multi-tenant policies and data quality checks for reliable, scalable deployment.

Qualifications

• Experience delivering a production data lake from end to end, storage layout, partitioning, tiered retention, access control.
• Deep knowledge of Spark (PySpark/Scala), including performance tuning and maintenance of Iceberg/Delta Lake tables.
• Hands-on experience with orchestration systems (Flyte, Kubeflow Pipelines, Airflow, or Prefect) in production.

Responsibilities

• Define architectural direction for training data flow, governance, and privacy across the AI Platform.
• Own data pipelines and services, ensuring tenant isolation and PII-absence gates.
• Lead multi-tenant data architecture, data quality gates, and model signal upsert processes.

Skills

Principal SWE

Platform standards

Data lake ownership

Spark PySpark

Beam/Dataflow

Flyte orchestration

Multi-tenant architecture

Feature store operations

Vector databases

RAG fundamentals

gRPC / TLS

Education

Bachelor's degree in CS/Engineering

Tools

Dataproc

Iceberg

Protobuf/Avro

Dataflow

Flyte

Kubeflow Pipelines

Airflow

Prefect

Pgvector

Qdrant

Job description

Saviynt seeks a Principal SWE to steer the AI Platform's data architecture, ensuring secure, scalable data flows from raw inputs to model signals. You will define standards for ML data pipelines, governance, and PII safeguards while maintaining cross-team collaboration across a large SaaS platform.

You will own Spark-based processing, streaming with Dataflow, and orchestration with Flyte and Kubeflow, plus multi-tenant policies and data quality checks for reliable, scalable deployment.