At FinBox, we're on the lookout for exceptional folks who are all about innovation and impact. If you're excited to shake things up in the banking and financial services world, keep reading!

Responsibilities

- Turn alternative data into financial identity. Convert consented SMS and app metadata signals into rich, accurate financial profiles and a customer 360 for people traditional credit systems can't see.
- High-accuracy income and fraud frameworks across borrower types. Build robust income-estimation and fraud-detection frameworks that hold up for salaried, self-employed, and MSME borrowers, each with a very different data footprint.
- Multi-modal underwriting intelligence with triangulation. Fuse GST, ITR, bank statements, and bureau data into a single reliable view, resolving conflicts across sources to produce underwriting-grade signals.

- Propensity, credit scores, and the Loan Recommendation Engine (LRE). Build propensity and credit-risk models and an LRE that uses superior algorithms to match the right product to the right customer, measurably lifting the likelihood a customer takes and repays a loan.
- Cost-optimized engagement. Optimize channel economics, e. g., WhatsApp outreach, so the right customers are reached with the right nudge at the right time at the lowest cost.

Underwriting Deep Dive, Where The Domain Depth Lives

- This is a role for someone who is both a quantitative modeler and a genuine underwriting mind, able to express credit judgment in code and lift the craft of the team around them.
- MSME, LAP, and micro-LAP underwriting. Own the models and decisioning behind secured MSME lending, where borrowers are traders, self-employed professionals, and small businesses with little or no formal income proof.

Build cash-flow-based income and affordability models for micro-LAP and LAP, translate fair-market vs. distress-sale valuation into a risk-adjusted LTV, and encode underwriting policy, eligibility,



obligations/FOIR, exposure caps, etc. directly in code. Shape how CAM (Credit Assessment Memo) generation is automated and be the quantitative underwriting authority who can guide credit teams on how MSME and property-backed risk is really assessed.

- Underwriting on banking data. Build the cash-flow engine that reads a bank statement the way a seasoned credit officer wouldaverage monthly balance, credit/debit turnover, EMI and obligation load, bounce/NSF patterns, and income stability, and turns it into income surrogates and affordability for borrowers with no formal documentation. Model the working-capital cycle (the timing gap between supplier payments and customer receipts) to catch receivables stress early, handle multi-account and group-entity structures, and reconcile banking turnover against GST.

Bank-statement-derived signals routinely outperform anything on the application form, so your job is to make them production-grade.

- Underwriting on bureau data. Build risk models that blend the commercial bureau (CIBIL MSME Rank / CMR) with the promoter's consumer bureau because a strong personal score can mask a stressed business, and vice versa. Mine the DPD grid, live obligations and utilization, inquiry velocity, and credit vintage into default and affordability signals; compute FOIR from trade lines; and design how thin-file / recent-to-credit MSMEs are scored, precisely where bureau depth runs out and FinBox's alternate data takes over.

Requirements

- > 4 years total experience, with at least 3 years in India's financial-services / fintech / lending domain.
- A builder, not just an analyst: you've shipped models (risk, fraud, propensity, or similar) into production.
- Fluent in Indian credit data: GST, ITR, bank statements, and bureau and how lending decisions are actually made.
- Quantitative underwriting depth, especially MSME: able to translate underwriting into code and mentor others.

This job was posted by Nikila Belle from FinBox.

📌 Senior Data Scientist (Bengaluru)
🏢 FinBox
📍 Bengaluru