Job description

Job Role: Senior Data Scientist – Credit Models & Applied AI

Experience: 5+ Years

Location: Mumbai

About the role

• Own end-to-end development of credit scorecards and decision analytics across bureau, platform, behavioural and portfolio data – from population and target definition through deployment and monitoring.

• Work with Credit, Risk, Underwriting and Technology to convert model outputs into grades, approval treatment, limit, pricing, tenure and reason codes. Applied AI is a selective secondary capability for document, evidence and analytical assistance – not autonomous financial decisioning.

Key Responsibilities

• Define development populations, observation/performance windows and targets using portfolio maturity, vintage, roll-rate and business context; benchmark existing scores before recommending a new model or recalibration.

• Clean and profile data, engineer interpretable features, prevent leakage and build explainable benchmark and challenger models across bureau, platform, repayment and other approved data.

• Complete champion-challenger selection and validation using KS, Gini/AUC, calibration, stability/PSI, out-of-time and segment performance; create score scaling, grades, reasons, limitations and model documentation.

• Translate selected models into policy/BRE treatment, approval/referral/rejection, limit, pricing and tenure logic; prepare deployment artefacts, golden cases, API/UAT evidence and productionmonitoring requirements.

• Develop EWS, collections, fraud/trust, propensity and portfolio analytics, and selectively support grounded NLP/LLM use cases such as document extraction, evidence retrieval and internal risk summaries.

Core Competencies

• Strong statistical discipline combined with practical credit judgement – able to distinguish predictive lift from leakage, instability or weak business meaning.

• Hands-on ownership mindset: comfortable coding, challenging data, presenting decisions and following models through production monitoring.

• Clear communicator who can explain model behaviour, limitations and business impact to Credit, Underwriting, Technology, management and assurance teams.

• Understanding of detailed data statistics methods like regression, time series, sampling theory, hypothesis testing etc.

Must-Have Requirements

• 5-8 years of hands-on Data Science / statistical modelling experience, including at least 3 years in lending, credit risk, underwriting or closely related BFSI analytics; strong Python and SQL are mandatory.

• Personally built at least one credit scorecard or underwriting model end-to-end, including population/target design, feature engineering, validation, calibration, deployment support and monitoring.

• Strong understanding of bureau and lending data, model governance, explain ability, reason codes and production implementation through policy/BRE, APIs or decision engines.

• Exposure to logistic scorecards and ML methods such as • XGBoost/CatBoost/LightGBM, along with cloud, MLflow, APIs, Git and MLOps practices.

• B.e/B.Tech/B.Stat

Good-to-Have (Optional)

• Experience in MSME, embedded finance, line-of-credit, working-capital, collections, fraud or earlywarning analytics.

• Practical exposure to NLP/LLM, RAG or document-intelligence use cases with grounding, evaluation, privacy controls and mandatory human review.

Ideal Candidate Profile

• A credit modeller first: technically strong, commercially aware and able to translate statistics into defensible lending treatment.

• Has worked with imperfect real-world data and can make transparent decisions on exclusions, missingness, stability, segments and implementation trade-offs.

• Uses Applied AI selectively where it improves analysis or productivity, without over-engineering the role or weakening model and policy governance.