Skip to content

FinTech Prompt Engineering project (ZeTheta Internship). Uses FPF and Multi-Layer Strategy to generate a secure Robo-Advisory microservices architecture and core MPT Python code with full validation.

License

Notifications You must be signed in to change notification settings

alearisteguieta/Prompt-Engineering-FinTech-FPF

🏦 Personal Wealth Management LLM Orchestration (FinTech Prompt Engineering)

License: MIT Technology: Python Methodology: FPF Status: Internship Project

📌 Executive Summary

  • This repository contains my first Prompt Engineering project, developed during my internship at ZeTheta Algorithm Private Limited.
  • The project demonstrates the application of advanced prompt engineering methodologies to design a personal wealth management system that integrates
  • multi-bank account aggregation (Mint) with automated investment advice (Betterment)

🔎 Repository Scope

  • This repository is a condensed, portfolio-style version of the full "Personal Wealth Management App (Mint + Betterment)" project. Its goal is to showcase the architecture, prompt methodology (FPF + Multi-Layer), and key artifacts in a concise manner.
  • Contents: curated excerpts, examples, and essential specifications.

  • Full Content: detailed documentation, full prompts, end-to-end traceability, tests, and results.

  • To review the full project documentation, visit: [Full Documentation]

Notes:

  • This repository does not include sensitive data or credentials.
  • Some sections have been simplified for space, privacy, and pedagogical clarity.

🎯 Objectives

  • Apply the Breakdown Problem Structure to decompose a FinTech system into manageable components.
  • Implement the Financial Prompt Framework (FPF) and the Multilayer Prompt Strategy (MPF) to ensure compliance, scalability, and traceability.
  • Deliver a complete and documented repository as a professional and educational reference.

🧩 Methodology

The project follows a strict prompt architecture sequence

  1. Meta Prompt
  2. Financial Prompt Framework (FPF)
  3. Multilayer Prompt Strategy (MPF)

Breakdown Problem Structure

  • Step 1 – Architecture: secure design, APIs, DB schemas.
  • Step 2 – Backend Logic: categorization engine, budget, investment algorithms.
  • Step 3 – Frontend/UI: financial data visualization, mobile design.
  • Step 4 – Integration/Security: MFA, encryption, fraud detection.

Multilayer Prompt Strategy

  • Layer 1 – Strategic Prompts: Define architecture, schemas, APIs.
  • Layer 2 – Development Prompts: Generate MPT, Categorization, Tax-Loss modules.
  • Layer 3 – Refinement Prompts: Implement security, generate tests, configure deployment.

📚 Documentation


🧪 Testing & Iterations

Validation and iterative improvements are documented in:


💡 Example: Layer 1 - Strategic Architecture Prompt

This is an excerpt of the initial, high-leverage prompt used to define the project's foundation. It mandates strict technical constraints and verifiable output formats.

fintech_architecture_prompt = """
**ROLE:** Act as a Senior FinTech Architect, expert in building high-performance Robo-Advisory systems (Trading/ML) and strict regulatory compliance (PCI DSS, Zero Trust, OAuth 2.0).

**OBJECTIVE:** Design a complete, secure, and scalable Microservices architecture for a Personal Wealth Management application (Mint + Betterment).

## STRATEGIC REQUIREMENTS (FPF)
1. **Security/Availability Metric (SLA):** The architecture must guarantee **99.99% uptime** for the investment engine.
2. **Performance Metric (Latency):** The Expense Categorization Engine (ML) must process 95% of transactions in **less than 500 milliseconds**.
3. **Zero Trust Security:** Implement network segmentation, Hashicorp Vault/AWS Secrets Manager, and **AES-256** encryption for data at rest.

## TECHNICAL REQUIREMENTS AND OUTPUT
**Technologies:** Python (Backend/ML), Flask/Django, PostgreSQL, Redis, Plaid API.
**Output Format:** Document structured in Markdown (NO code).

The document must include:
1. Descriptive Microservices Architecture Diagram.
2. Key Database Schemas (User, Transaction, Recommendation), detailing hashing (Bcrypt) and encrypted fields (AES-256).
...
"""

📂 Repository Structure

📂 Prompt-Engineering-Project
┣ 📜 README.md
┣ 📂 docs
┃ ┣ 📜 introduction.md
┃ ┣ 📜 breakdown_problem_structure.md
┃ ┗ 📜 multilayer_prompt_strategy.md
┣ 📂 examples
┃ ┣ 📜 prompt_case1.md
┃ ┣ 📜 prompt_case2.md
┃ ┗ 📜 generated_outputs.md
┣ 📂 testing_validation
┃ ┣ 📜 test_plan.md
┃ ┣ 📜 validation_results.md
┃ ┗ 📜 unit_tests.py
┣ 📂 iterations
┃ ┣ 📜 v1_initial.md
┃ ┣ 📜 v2_refined.md
┃ ┗ 📜 scalability_notes.md
┗ 📂 assets
  ┗ 📜 diagrams.png
┣ 📜 CONTRIBUTING.md
┃ 📜 SECURITY.md
┃ 📜 CODE_OF_CONDUCT.md
┃ 📜 PULL_REQUEST_TEMPLATE.md

About

FinTech Prompt Engineering project (ZeTheta Internship). Uses FPF and Multi-Layer Strategy to generate a secure Robo-Advisory microservices architecture and core MPT Python code with full validation.

Topics

Resources

License

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published