| Layer | Tool | Why | Minimum You Must Learn |
|---|---|---|---|
| Data store | PostgreSQL | Free, real RDBMS, widely used in risk infra | Schemas, indexes, joins, window fns, materialized views |
| Data wrangling | Python (pandas, numpy) | Glue between raw data + analytics | ETL from CSV/API → Postgres; feature engineering; stats |
| Modeling | Python (scikit-learn / statsmodels) | Scorecards, regressions, simple PD models | Logistic regression, ROC/AUC, calibration |
| Reporting / Ops | Excel + Power Query | Everyone still uses Excel in risk. Period. | Connect to Postgres, refresh dashboards, what-if |
| Visualization | Power BI or Tableau Public | Exec-facing story; risk heatmaps | Data model, slicers, filters, KPI cards |
| Workflow | GitHub | Proof of work + versioned notebooks | README-first structure; push weekly |
| Knowledge hub | Notion | Index your projects w/ short case writeups | 1 page per project: Problem → Data → Method → Metric → Screenshot |
| Signal | Recruiters search it; network sees progress | 2 posts/week: build log + insight takeaway |
Alright, you asked for an undeniable, full-stack Python + SQL + Excel + Analytics + Finance/Risk skill-build that makes you dangerous in conversations, credible in interviews, and impossible to ignore on LinkedIn. No fluff. No “Day 5: Learn WHERE clause.” This is integrated, portfolio-driven, signal-heavy output that shows you can do real work that matters to risk teams.
In ~12 weeks (or faster, if you sprint): build a public, version-controlled portfolio of finance + risk analytics projects across SQL, Python, Excel, and viz, each one producing insights you can talk about like a pro. By the end:
| Layer | Tool | Why | Minimum You Must Learn |
|---|---|---|---|
| Data store | PostgreSQL | Free, real RDBMS, widely used in risk infra | Schemas, indexes, joins, window fns, materialized views |
| Data wrangling | Python (pandas, numpy) | Glue between raw data + analytics | ETL from CSV/API → Postgres; feature engineering; stats |
| Modeling | Python (scikit-learn / statsmodels) | Scorecards, regressions, simple PD models | Logistic regression, ROC/AUC, calibration |
| Reporting / Ops | Excel + Power Query | Everyone still uses Excel in risk. Period. | Connect to Postgres, refresh dashboards, what-if |
| Visualization | Power BI or Tableau Public | Exec-facing story; risk heatmaps | Data model, slicers, filters, KPI cards |
| Workflow | GitHub | Proof of work + versioned notebooks | README-first structure; push weekly |
| Knowledge hub | Notion | Index your projects w/ short case writeups | 1 page per project: Problem → Data → Method → Metric → Screenshot |
| Signal | Recruiters search it; network sees progress | 2 posts/week: build log + insight takeaway |
Each project builds on prior work. Reuse data + code. Ship fast, refine later.
Outputs:
udipt-risk-lab.