Applied Agentic AI Engineering for Banking Professionals

Why Choose Garranto Academy for Applied Agentic AI Engineering for Banking Professionals?

Garranto Academy delivers a hands-on, production-focused bootcamp designed for banking technology teams and AI professionals. Participants develop practical engineering skills through real-world banking workflows, regulatory scenarios, and the BankBot360 project, progressing from Python fundamentals to building and deploying production-ready, auditable multi-agent AI banking systems.

Course Overview:

Applied Agentic AI Engineering for Banking Professionals is a 120-hour, 30-day practical bootcamp that prepares participants to design, build, and deploy production-ready AI systems for regulated banking environments. Starting with Python fundamentals and large language model integration, the programme progresses through LangChain, Retrieval-Augmented Generation, vector databases, autonomous agents, and multi-agent orchestration. Participants develop BankBot360 throughout the course, applying their skills to banking workflows such as regulatory compliance, fraud investigation, credit assessment, and KYC/AML processing. They also build MCP integrations, secure FastAPI services, PostgreSQL audit trails, and Streamlit dashboards, before containerising the solution with Docker and configuring CI/CD. Combining 60% hands-on labs with 40% concepts and frameworks, the programme concludes with a live demonstration of the integrated system for banking teams.

What You'll Learn in Our Applied Agentic AI Engineering for Banking Professionals Course?

Course Objectives:

Upon successful completion of this course, learners will be able to:

• ✓ Develop production-quality Python applications for banking workflows.

• ✓ Integrate OpenAI and Claude APIs reliably.

• ✓ Build RAG pipelines over banking regulatory documents.

• ✓ Manage Chroma and Pinecone vector databases.

• ✓ Expose banking capabilities through MCP servers.

• ✓ Build secure FastAPI services with audit logging.

• ✓ Deploy containerised AI systems using Docker.

• ✓ Apply governance, evaluation, and human oversight controls.

Prerequisites

• ✓ No prior programming experience is required

• ✓ Basic arithmetic and percentages are sufficient

• ✓ General banking operations familiarity is helpful

• ✓ OpenAI, Anthropic, and Pinecone accounts required

Course Outlines:

Module 1: Python Foundations for AI (Days 01–05)

• ✓Day 1: Setup and Core Syntax – Explore Python environment setup, variables, data types, f-strings, conditionals, and loops.

• ✓Day 2: Data Structures and Functions – Understand lists, dictionaries, sets, tuples, list comprehensions, functions, arguments, and lambda expressions.

• ✓Day 3–5: Apply object-oriented programming, file handling, exception management, pandas, REST APIs, and asynchronous programming.

Module 2: LLM APIs and Prompt Engineering (Days 06–09)

• ✓Day 6: Calling LLM APIs – Integrate OpenAI and Anthropic APIs, understand message roles, token counting, and cost estimation.

• ✓Day 7: Prompt Engineering – Explore zero-shot, one-shot, few-shot prompting, Chain-of-Thought, CO-STAR, and RAISE frameworks.

• ✓Day 8–9: Implement structured outputs using Pydantic, JSON mode, streaming, asynchronous requests, and batch processing.

Module 3: LangChain and RAG Pipelines (Days 10–14)

• ✓Day 10: LangChain Architecture and LCEL – Explore LangChain components, runnable sequences, prompt templates, and LangSmith tracing.

• ✓Day 11–12: Implement conversation memory, document loaders, text splitting, embeddings, and persistent storage.

• ✓Day 13–14: Build complete RAG pipelines, source citations, conversational retrieval, and metadata filtering.

Module 4: Vector Databases – Chroma to Pinecone (Days 15–17)

• ✓Day 15: Chroma – Local Vector Database – Explore database operations, document management, metadata filtering, and LangChain integration.

• ✓Day 16: Pinecone – Production Vector Database – Configure serverless indexes, metadata filters, batch upserts, and namespace isolation.

• ✓Day 17: Migration and Abstraction – Implement vector database migration and reusable storage using the VectorStore factory pattern.

Module 5: LangGraph and Agentic Workflows (Days 18–21)

• ✓Day 18: LangGraph Fundamentals – Explore ReAct patterns, StateGraph, conditional routing, and checkpointing.

• ✓Day 19–20: Develop banking tools, integrate transaction analysis, and build autonomous fraud investigation agents.

• ✓Day 21: Human-in-the-Loop and Subgraphs – Implement officer approval workflows, checkpoints, and reusable agent components.

Module 6: Model Context Protocol (MCP) (Days 22–24)

• ✓Day 22: MCP Architecture – Understand MCP hosts, clients, servers, tools, resources, prompts, and transport mechanisms.

• ✓Day 23: Building the BankBot360 MCP Server – Develop banking tools, regulatory resources, prompt templates, and validate integrations.

• ✓Day 24: MCP Clients and LangGraph Integration – Connect MCP clients, integrate banking tools, and configure remote communication.

Module 7: CrewAI Multi-Agent Orchestration (Days 25–27)

• ✓Day 25: CrewAI Fundamentals – Explore AI agents, roles, goals, tasks, crews, and sequential workflows.

• ✓Day 26: Hierarchical Crews and MCP Integration – Implement hierarchical orchestration, MCP tools, and parallel processing.

• ✓Day 27: Full KYC Pipeline – Build a five-agent KYC/AML workflow with document validation, sanctions screening, risk scoring, and compliance reporting.

Module 8: FastAPI and PostgreSQL Backend (Days 28–29)

• ✓Day 28: FastAPI AI Endpoints – Develop REST endpoints, request/response schemas, AI routers, and streaming services.

• ✓Day 29: PostgreSQL, Authentication, and Testing – Implement database integration, audit logging, JWT authentication, rate limiting, and automated testing.

• ✓Integrate banking AI components into a secure backend with traceable operations.

Module 9: Streamlit UI, DevOps, and Capstone (Day 30)

• ✓Day 30: UI, Docker, and CI/CD – Build Streamlit dashboards, containerise applications using Docker, and configure GitHub Actions.

• ✓Develop compliance chat, fraud investigation, and KYC review interfaces.

• ✓Present the complete BankBot360 system through a live end-to-end demonstration.

Course Outcomes:

Upon completing the "Applied Agentic AI Engineering for Banking Professionals" course, participants will:

• ✓ Integrate multiple LLM providers with structured outputs.

• ✓ Build evaluated RAG pipelines with regulatory citations.

• ✓ Manage vector databases with tenant-level data isolation.

• ✓ Develop autonomous agents with human approval checkpoints.

• ✓ Expose banking tools through production-ready MCP servers.

• ✓ Orchestrate five specialist agents for KYC workflows.

• ✓ Deliver secure APIs with authentication and audit trails.

• ✓ Deploy containerised applications with automated CI/CD pipelines.

• ✓ Apply governance controls for regulated banking AI systems.

• ✓ Investigate suspicious transactions using agent-based workflows.

Key Benefits of Applied Agentic AI Engineering for Banking Professionals:

Garranto Academy equips banking professionals with practical AI engineering skills through the BankBot360 project, covering compliance, fraud detection, and KYC/AML automation. Participants develop expertise in multi-agent orchestration, secure integration, and production-ready AI deployment.

How Applied Agentic AI Engineering Can Transform Banking Operations?

Agentic AI transforms banking operations by streamlining regulatory compliance, fraud investigation, and customer onboarding. BankBot360 combines intelligent RAG assistance and multi-agent workflows with human approvals, audit trails, and governance controls to support secure, transparent, and efficient banking operations.