Credit Data Engineer
at Klarna
Posted 21 hours ago
No clicks
- Compensation
- Not specified
- City
- Not specified
- Country
- Not specified
Currency: Not specified
Build and own global underwriting (UW) data tables and pipelines to support scoring, real-time decisioning, monitoring, and regulatory reporting. Design agent- and human-friendly schemas, canonical events, consistent IDs, metric definitions, and machine-readable data contracts with clear SLAs for freshness, completeness, accuracy and lineage. Implement and operate batch and streaming pipelines (PySpark, Airflow, Kafka, Redshift on AWS), plus observability, alerts, audits and incident reviews. Collaborate closely with credit, policy, modeling, treasury and finance teams to deliver production-grade data products for underwriting and reporting.
What you’ll do
Own the global UW tables (canonical facts/dimensions for applications, decisions, features, repayments, delinquency) with clear SLAs for freshness, completeness, accuracy, and data lineage.
Design for AI-agents and humans: consistent IDs, canonical events, explicit metric definitions, rich metadata (schemas, data dictionaries), and machine-readable data contracts.
Build & run pipelines (batch + streaming) that feed UW scoring, real-time decisioning, monitoring, and underwriting optimization.
Instrument quality & observability (alerts, audits, reconciliation, backfills) and drive incident/root-cause reviews.
Partner closely with Credit Portfolio Management, Policy teams, Modeling teams, and treasury and finance teams to land features for RUE and consumer-centric models, plus regulatory and management reporting.
Tech stack (what we use)
Languages: SQL, PySpark, Python
Frameworks: Apache Airflow, AWS Glue, Kafka, Redshift
Cloud & DevOps: AWS (S3, Lambda, CloudWatch, SNS/SQS, Kinesis), Terraform; Git; CI/CD
What you’ll bring
Proven ownership of mission-critical data products (batch + streaming).
Data modeling, schema evolution, data contracts, and strong observability chops.
Familiarity with AI/agent patterns (agent-friendly schemas/endpoints, embeddings/vector search).
