THE PIPELINE

From raw data to credit decision in milliseconds.

A walkthrough of the TrueScore scoring architecture across data ingestion, feature engineering, model scoring, explainability, and delivery.

Step 1 - Data Ingestion

Scaffold coverage includes bureau pulls, AA consent and fetch, alternative data via Perfios / Finbox, DigiLocker verification, GSTN fetches, and normalisation before scoring.

Step 2 - Feature Engineering

Individual pipeline

FeatureSourceType
FOIRBank statement / AARatio
DTIBureau + income docsRatio
Repayment consistency scoreBureau historyDerived
Cash flow regularityAA / bank stmtDerived
Bureau vintageCIBIL / ExperianRaw
Enquiry intensity (90d)BureauCount

Business pipeline

FeatureSourceType
DSCRBank stmt / PerfiosRatio
GST turnover trend (12m)GSTNTrend
Filing compliance rateGSTNRate
Promoter credit scoreBureauRaw
Charge or lien on assetsMCA21Flag
Sector risk overlayRBI sectoral dataScore

Step 3 - LightGBM Scoring Engine

Prompt 02 frames LightGBM as the preferred model family for speed, categorical feature handling, and better explainability tradeoffs than deeper black-box alternatives.

ModelAUC-ROCKS StatGini
Logistic Regression (baseline)0.740.380.48
Random Forest0.790.440.58
XGBoost0.820.480.64
TrueScore (LightGBM)0.910.620.82

Step 4 - SHAP Explainability

Every score should map back to human-readable drivers so a lender can explain adverse action without treating interpretability as an afterthought.

Step 5 - Delivery

REST API, webhook delivery, batch scoring, and dashboard consumption should all resolve from the same scoring contract and audit trail.