feat: the UdyamLinc ML layer — scoring, explanation and quantile forecast

Brings the Python service under version control for the first time. It has
been running and serving the demo since 23 August; until now it existed only
on one developer's disk.

Contents:

  training/build_dataset.py   point-in-time join over the feature store, with
                              leakage assertions. dscr_p10 and dscr_p50 are
                              excluded deliberately — both are derived from
                              the forecast, which is derived from the
                              instalment, so the model would read its own
                              answer.
  training/train_stress.py    LightGBM vs XGBoost vs two single-feature
                              baselines, expanding-window validation over
                              as-of dates, publishes to the model_metric table
  training/train_forecast.py  three quantile heads (P10/P50/P90), horizon as a
                              feature rather than one model per horizon
  serving/explain.py          SHAP TreeExplainer folded into the five risk
                              families the console already speaks. The family
                              map mirrors RuleBasedScoringStrategy in the API
                              repo; where they disagree the rule engine wins.
  serving/app.py              /score, /score/batch, /forecast, /health, /model

The trained models, their metrics and the training dataset are committed
alongside the code. Without them the published AUC of 0.760 is a claim nobody
else can check; at ~2.2 MB together they are worth the space.

Measured, on 5,523 rows / 263 enterprises / 242 positives / 100% synthetic
labels: LightGBM AUC-ROC 0.7599, and the emi_to_inflow baseline 0.7685 on the
same folds. The baseline wins. That is recorded rather than tuned away — on
generated labels, beating it would mostly prove the model had learned the
generator.

The quantile forecast beats the deterministic strategy on all three quantiles
(pinball loss, 12-33% better) but flags every borrower as breaching, so the
deployment runs udyam.forecast.engine=DETERMINISTIC and this model is held
back pending calibration against real repayment outcomes.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

data/manifest.json

@@ -0,0 +1,55 @@
+{
+  "rows": 5523,
+  "enterprises": 263,
+  "as_of_dates": 21,
+  "date_min": "2024-08-31",
+  "date_max": "2026-04-30",
+  "positives": 242,
+  "positive_rate": 0.0438,
+  "features": [
+    "feed_cost_change",
+    "output_price_change",
+    "margin_squeeze",
+    "price_z_score",
+    "rainfall_anomaly",
+    "heat_anomaly",
+    "cash_flow_change",
+    "emi_delay",
+    "inventory_days",
+    "upi_inflow_change",
+    "feed_staleness_days",
+    "days_since_last_entry",
+    "emi_to_inflow"
+  ],
+  "excluded": {
+    "dscr_p10": "reads forecast_point with no as-of filter \u2014 present leaks into past",
+    "dscr_p50": "same"
+  },
+  "horizon_months": 3,
+  "dpd_threshold": 30,
+  "label_source_split": {
+    "SEEDED": 273,
+    "SEEDED_HISTORY": 6039
+  },
+  "synthetic_label_share": 1.0,
+  "feature_coverage": {
+    "feed_cost_change": 84.1,
+    "output_price_change": 97.7,
+    "margin_squeeze": 83.4,
+    "price_z_score": 90.3,
+    "rainfall_anomaly": 100.0,
+    "heat_anomaly": 100.0,
+    "cash_flow_change": 100.0,
+    "emi_delay": 100.0,
+    "inventory_days": 100.0,
+    "upi_inflow_change": 0.0,
+    "feed_staleness_days": 100.0,
+    "days_since_last_entry": 100.0,
+    "emi_to_inflow": 100.0
+  },
+  "caveats": [
+    "Feature rows are reconstructed by fact date, not knowledge date \u2014 the archive was ingested in one pass, so a past row uses a value a live run of that date would not yet have had.",
+    "Labels are generated DPD history (SEEDED_HISTORY). Metrics computed on this dataset measure the generator as much as the borrowers, and must be reported as computed on synthetic history.",
+    "dscr_p10 and dscr_p50 are excluded: both read the current forecast at every as-of date."
+  ]
+}