Financer Intelligence

Banking Analytics Workbench

Financer Intelligence is a portfolio project designed around the work of an Associate Data Analyst in financial services. It turns a generic personal-finance app into a decision-support workbench for a synthetic retail-banking portfolio.

The project is deliberately transparent: it uses synthetic data only, never presents itself as an IDFC FIRST Bank product, and labels every dashboard state accordingly.

What It Demonstrates

Analyst capability	Evidence in the project
Business framing	Deposit growth, customer retention, lending stress, and action owners are presented as business decisions.
KPI reporting	Filterable active-customer, deposit-book, disbursal, and early-warning-account metrics.
Segmentation	Region, customer segment, product family, and reporting-period views update the analysis.
Risk monitoring	An explainable early-warning queue pairs risk score, trigger, exposure, and recommended action.
Data quality	Coverage, exception counts, check status, and accountable data owners appear beside decision metrics.
Product thinking	A React workbench, FastAPI data contract, responsive charts, and JSON snapshot export work end to end.
SQL reasoning	Reusable analytical SQL patterns are included in financer-backend/sql/portfolio_analysis.sql.
Decision modelling	A validated Scenario Lab makes assumptions, risk exposure, experiment design, and guardrails inspectable.
Engineering quality	GitHub Actions runs frontend lint/build and backend tests on every pull request and update to main.

Project Walkthrough

1. Open the landing page and select Explore workbench.
2. Change the portfolio slice by region, customer segment, product, or reporting period.
3. Read the portfolio trend alongside product performance and the risk distribution.
4. Review the decision queue and the early-warning watchlist.
5. Check data-quality controls before acting on the metrics.
6. Export the active synthetic snapshot to demonstrate a reproducible reporting workflow.
7. Open Scenario Lab to translate a retention or assisted-outreach hypothesis into a testable experiment.

Architecture

Synthetic banking model
        │
        ├── FastAPI analytics contracts
        │     ├── /analytics/overview
        │     ├── /analytics/metrics
        │     ├── /analytics/lineage
        │     └── /analytics/scenario
        │
        └── React analytics workbench
                 ├── KPI and trend charts
                 ├── segmentation and product tables
                 ├── risk action queue
                 └── data quality controls

Stack

• Frontend: React 19, Vite, React Router, Tailwind CSS, Recharts, Framer Motion
• Backend: FastAPI, Pydantic, in-memory/optional MongoDB data service
• Analysis: deterministic synthetic portfolio model plus SQL query examples

Run Locally

Frontend

cd financer
npm install
npm run dev

Backend

cd financer-backend
py -m pip install -r requirements.txt
py -m uvicorn main:app --reload

The dashboard works immediately with its local synthetic fallback. When the backend is running, it also fetches the same snapshot through GET /analytics/overview.

Verify

cd financer
npm run lint
npm run build

cd financer-backend
py -m pytest test.py -q

Interview Notes

Use the project to explain a clear analytical loop:

1. Define a decision and its success metric.
2. Segment the portfolio to find the material driver.
3. Validate data completeness and freshness.
4. Convert the result into an owner, action, and priority.
5. Track whether the action improves the KPI in the next reporting period.

The synthetic data is intentionally simple and deterministic so the analysis is easy to inspect. In a production setting, replace it with governed warehouse tables, role-based access, audited transformations, and approved monitoring thresholds.

Evidence Pack

• docs/ANALYTICS_CASE_STUDY.md explains the business problem, metric logic, analysis plan, and responsible limitations.
• docs/INTERVIEW_DEMO_PLAYBOOK.md provides a 90-second story, a 6-minute demo route, credible resume bullets, and likely technical follow-ups.