Design and own the feature-store architecture that credit decisioning will depend on. You will own the architecture and inner workings of agentic harnesses and loop/graph agent workflows bound to the feature store: feature engineering, point-in-time validation, drift and data-quality observability.
Locations (2)
Warsaw, Mazovia, Poland
Kraków, Poland
Responsibilities
Design and own the feature-store architecture for credit decisioning
Set pipeline patterns, modelling standards, and lineage approach
Lead highest-complexity, highest-risk data initiatives
Design the agent loop (context, tools, evals, guardrails) for reliable agentic engineering
Own architecture and inner workings of agentic harnesses and loop/graph agent workflows
Requirements
Nice to have: Feature-store or ML data serving in production
Nice to have: Payments experience
Nice to have: Fintech experience
Nice to have: Lending data experience
Nice to have: Streaming (Kafka/Structured Streaming)
Design, build, and operate batch and streaming data pipelines on the lakehouse (Databricks / Spark) that ingest, validate, and model data for credit origination, underwriting, and portfolio reporting.
Locations (2)
Warsaw, Mazovia, Poland
Cracow, Poland
Responsibilities
Design, build, and operate batch and streaming data pipelines on the lakehouse (Databricks / Spark) that ingest, validate, and model data for credit origination, underwriting, and portfolio reporting
Build and maintain the feature store that BNPL Origination and MCA Underwriting consume, guaranteeing point-in-time correctness, freshness SLAs, and reproducibility for every served feature
Build and operate AI-first / agentic processes for the team's pipelines — designing the agent loop (context, tools, evals, guardrails) so agentic delivery is reliable, observable, and cost-controlled
Model financial and behavioural data into governed, documented datasets with clear ownership, lineage, and access controls aligned to GDPR and internal data governance
Embed data-quality and observability as first-class deliverables: freshness, completeness, distribution, and schema-drift checks with alerting