Scorecard Developer (Machine Learning Specialist) - Bengaluru

Role purpose

As a Scorecard Developer, you'll develop and maintain credit scoring components and associated calibrations that support approval and risk strategies across products and markets. You'll focus on building high-quality features, ensuring scores are stable and explainable, and delivering robust PD-to-bad-rate calibrations that translate model outputs into decision-ready risk measures.

Key responsibilities

·        Develop and maintain scoring solutions and supporting artefacts used in credit decisioning (application and/or behavioural scoring, segmentation, risk signals).

·        Own feature engineering for scoring: create, test and document variables from bureau, application, transactional and repayment data; ensure stability, interpretability and data quality.

·        Contribute to model development and tuning using modern machine learning approaches where appropriate, ensuring outputs are robust, stable and suitable for decisioning.

·        Apply best-in-class machine learning practices for credit scoring, including disciplined hyperparameter optimisation, robust validation, and repeatable model selection workflows appropriate for production decisioning.

·        Define and maintain feature specifications for production (definitions, transformations, edge-case handling, missing value logic, consistency checks).

·        Produce PD / score calibrations to observed bad rates (overall and by segment), including calibration curves, stability tracking, and recalibration recommendations.

·        Support cut-off / limit strategy analysis using calibrated risk outputs (approval rate vs bad rate vs loss trade-offs).

·        Run ongoing monitoring: drift and stability of inputs/features, score distribution shifts, performance by segment and cohort/vintage, data pipeline health.

·        Partner with Engineering / Decisioning teams to operationalise scoring outputs and ensure reproducibility (versioning, back-testing, change control).

·        Maintain clear documentation suitable for internal review/audit (feature catalogue, calibration approach, monitoring packs, change logs).