CoolFace
Apppublic

NitinThapliyal/Banking-Advisory-Copilot

sourceHugging Facemitupdated 4mo agoView on Hugging Face
0likes
App README

Banking Support & Advisory

A minimal but complete capstone project demonstrating an industry-style agent workflow across:

  • —baseline agent design
  • —LLM integration
  • —retrieval
  • —tool use
  • —planning and memory
  • —adaptive behavior
  • —deployment readiness
  • —evaluation

Scenario

The app helps users learn about different banking transactions and loan inquiries:

  1. 1.Loan eligibility**
  2. 2.Banking Transactions**

Project structure

text
weekend-trip-copstone/
├── README.md                       # you are here
├── app.py                          # entry point for streamlit
├── agent_system.py                 # All Agentic code and logic
├── retrieval.py                    # RAG Pipeline
├── evaluation.py                   # Test/Evaluate system
├── config.py                       # Environment Variables
├── requirements.txt                # All required packages            
├── Dockerfile                      # Required for Deployment
├── .env                            # All required API keys
├── data/                           # Vector Store
│   └── knowledge_base/
├── docs/                           # Capstone Report
│   └── capstone_report.md

Core capabilities

  • —Baseline mode for Phase 2 evidence
  • —Smart mode for the final agent system
  • —one tool
  • —Tavily search
  • —Retrieval
  • —embeds local Banking transactions and loan details MS word documents
  • —Short-term memory
  • —keeps recent turns
  • —Adaptive behavior
  • —updates style/priority based on feedback buttons
  • —Evaluation
  • —compare baseline vs smart vs smart+retrieval

Local setup

1. Create environment

bash
python -m venv .venv

Activate it:

macOS / Linux

bash
source .venv/bin/activate

Windows

bash
.venv\Scripts\activate

2. Install dependencies

bash
pip install -U pip
pip install -r requirements.txt

3. Configure .env

Setup .env and fill in keys.

4. Run the app

bash
streamlit run app.py

5. Run evaluation

bash
python evaluation.py

This generates:

  • —docs/runtime_logs.jsonl
  • —docs/evaluation_results.csv

Deployment

This app is intended for Docker deployment on Hugging Face Spaces.

Evidence map

See docs/capstone_report.md for phase-wise evidence and explanations.

Suggested prompts

  • —What are different types of transactions available at bank?
  • —What are the different type of loans one can have?
  • —What is the eligibility for home loan?
  • —Are there any penalties for late payments?