The mission
Set the ML architecture and modelling standards that credit decisioning will depend on.
- Standards others copy. You set the modelling, validation, monitoring and responsible-AI / model-risk practices for credit-scoring and MCA merchant-underwriting models across the platform.
- Lifecycle governance. You own the end-to-end model lifecycle, from feature selection and training methodology through validation, champion/challenger testing, deployment and post-production monitoring for drift and stability.
- Built on the feature store. You partner with Data Engineering to define point-in-time-correct, production-grade features. You shape the feature-store roadmap around modelling needs.
- Regulatory defensibility by construction. You establish documentation, model cards and adverse-action reasoning standards.
- Agent-first delivery. You own the architecture of agentic harnesses and loop/graph agent workflows bound to the feature store: agentic model iteration (experiment, evaluate, retrain, promote) and agentic model monitoring and observability. You design the agent loops and graphs (context, tools, state, evals, guardrails) so they are reliable, observable and cost-controlled. You codify them as the reference patterns the function reuses.
- Raise the bar. You anchor the ML engineering team: you set technical direction, review designs and code, and mentor Senior and Mid ML Engineers.
What you will work with
The feature store, model registry and serving, MLflow and Databricks, agent orchestration graphs, eval harnesses, and model-monitoring and tracing tools. You will also work with Credit Risk and the engineering teams that integrate models into decisioning services.
You are:
- Someone who loves building ML systems that hold up in production and under audit.
- Hands-on first. You design and you build, with AI as a daily tool.
- Experienced, with 7+ years in machine learning engineering in production systems, and 3+ years in credit-scoring, risk modelling or decisioning ML on Spark/Databricks or an equivalent lakehouse.
- Expert in Python and ML frameworks (scikit-learn, XGBoost, PyTorch) (7+ years).
- Strong in credit-risk and underwriting model design: scorecards, gradient boosting, survival/behavioural models.
- Skilled in model validation methodology: back-testing, champion/challenger, population stability (PSI), calibration.
- Experienced in point-in-time-correct feature engineering, feature-store consumption patterns and MLOps tooling (MLflow, CI/CD for models).
- Fluent in explainability and responsible-AI techniques for regulated decisioning (SHAP, model cards, adverse-action reasoning), with working knowledge of regulatory model-risk frameworks.
- Experienced with agentic coding tools and loop and graph engineering (3+ years): agentic harnesses, context and tool design, evals, guardrails, iteration. You have deep knowledge of LLM behaviour and trade-offs.
- A Staff-level engineer who sets standards through influence, not authority, and mentors through code and design review.
- Fluent in English (B2+).
Bonus points for:
- FinTech, lending or payments ML
- Model-risk management or regulatory model validation
- MLOps or feature-store platform ownership
- Fraud-detection modelling (bust-out, first-party fraud)
- Real-time, low-latency model serving
- Board- or regulator-facing model documentation and audit experience
- Causal inference / uplift modelling for credit-policy experimentation
Why InPost?
- Architectural authority from day one. You set the standard.
- Greenfield, high stakes. Your design decisions last for years.
- A modern stack and a lean, AI-native team, with space to experiment with new technologies and bring them into production.
- A collaborative, fast-moving environment where autonomy and accountability go hand in hand.
- Hybrid, ideally one day a week in the Warsaw or Kraków office.
- We offer a B2B Contract.