E
Fintech · Claude AI · Coding Automation · DX · AX

AI Agent · RAG · Cloud자동화 아키텍처를 설계합니다

AI Agent · RAG · Cloud
We design automation architectures

AI Sec Ops-based Air-Gap environment deployment · End-to-end execution from PoC to operations

With 25 years of hands-on experience and 113 delivered projects, we take enterprise AX/DX from PoC through operational optimization.

Our public projects also cover financial analytics, quant backtesting, financial RAG, and accounting and disclosure automation.

AI Service Architecture

Agent · RAG · Cloud
With one execution architecture

LIVE
Enterprise AI
AX · DX · Auto
AI Agent

Work automation orchestration

We design agents that carry out real work through prompts, tool calls, approval flows, and system integrations.

RAG / Data

Knowledge-grounded answers

We collect, chunk, embed, retrieve, and rerank documents to produce trustworthy, evidence-based answers.

DevOps / MLOps

Deployment and operational optimization

We connect cloud architecture, CI/CD, model operations, and observability for reliable services.

01 Discover
02 Design
03 Deploy
04 Optimize
ON-PREMISES AI ARCHITECTURE

Build LLM · RAG · AI agents inside your network

Enterprise on-premises AI architecture

A reference architecture connecting models, internal knowledge, and business tools inside a private network.

01 · Air-gapped supply chain

Validate and importBring in approved images and models as files
Harbor + internal CAPrivate registry and TLS trust chain
Robot AccountLeast-privilege image pull for Kubernetes

02 · Internal AI platform

KubernetesDeploy Harbor images to an internal cluster
GPU + vLLMLLM inference API
NFS / model storageKeep model weights and documents on premises

03 · Business services

RAG pipelineIngest, chunk, embed, and retrieve documents
AI AgentRetrieval, internal tools, and approval flow
Internal usersGrounded answers and work results

Harbor, offline import, Kubernetes image pull, and vLLM deployment come from harbor-lab exercises. RAG and agents extend that foundation as a proposed solution.

Company Overview

An AI Agent Engineering Partner
turning enterprise DX · AX into working systems

With 25 years of experience across education, the public sector, and enterprises, EDUMGT designs services ready for real operations. We work from strategy through development, deployment, and ongoing optimization.

Experience
25Y+

Hands-on delivery across AI, cloud, and LMS

Delivery
113+

Delivered projects and operations-focused outcomes

Focus
End-to-End

From planning and development to deployment and operations

AI Agent Studio RAG · Vector DB · Workflow LMS · LCMS · SCORM · xAPI AWS · Azure · Cloud Native DX / AX Consulting

AI Architecture Map

Live
LLM Agent RAG Cloud Tool API Vector Search CI/CD MLOps
Strength

Our strengths include domain understanding of education, public, and corporate systems and the ability to execute quickly.

Outcome

It also provides screens, data flows, and operating models that users actually use.

AI × Business solution

Claude (Anthropic) APIs power coding automation and AI consulting that improve how organizations work.

FINTECH × AI · GITHUB PROJECTS

From financial data to useful AI

Explore EDUMGT's public GitHub projects in investment analysis, financial knowledge retrieval, and accounting and disclosure automation. Each implementation links to its source repository.

Financial AI agents & RAG

Build ReAct-style financial Q&A, Qdrant document search, evidence-based answers from learning materials, and interactive analysis views.

Accounting & disclosures

Connect receipt OCR, account classification, anomaly detection, cash-flow forecasting, OpenDART data, and financial PDF generation.

AI · cloud · education solutions

On-Prem LLM · RAG · Agent

On-premises LLM, RAG & AI agents

Connect private LLM inference, internal knowledge retrieval, and business agents in one enterprise operating model.

  • ›Approved image import → Harbor → Kubernetes deployment
  • ›Internal LLM inference on GPU with vLLM
  • ›Permission-aware RAG with source evidence
  • ›Business agents with approvals and audit trails
Explore the 3D architecture ↗
Claude Biz Consulting

AI business application consulting

We assess core workflows and implement practical productivity improvements with Claude AI.

  • ›Plan AI transformation of business processes (as-is → to-be)
  • ›Automation of report writing, data analysis, and decision support
  • ›Design and build a Claude-based in-house knowledge management system
  • ›AI governance, security policies, and staff capability training

AI learning experience platform (LMS/LCMS)

It supports SCORM · xAPI · LRS integration, automatic creation of lesson plans/quizzes, and learner behavior-based analysis.

  • ›Content auto-completion/tagging
  • ›Detect and report underperforming learners
  • ›PDF/Video processing pipeline

AI Agent & Chatbot

Intelligent guidance and knowledge management automation, including in-house document-based RAG, multilingual, and voice/video response.

  • ›Teams · Slack · KakaoTalk integration
  • ›Sensitive information masking and audit log
  • ›Continuous learning/feedback loop

Data/Cloud Platform

Design data lake/warehouse, ETL/ELT, and real-time streaming environments.

  • ›Glue · Athena · Redshift
  • ›Step Functions · Lambda · MWAA
  • ›Security/Cost Optimization Architecture

Consulting & transformation

We provide a step-by-step roadmap from legacy analysis → modernization → modularization → operation.

  • ›PoC → MVP → operational adoption
  • ›Performance, security, quality inspection
  • ›Training to strengthen internal capabilities
Live AI Architecture

AI module linked flow

Select a scenario to see the actual data flow and how the modules work together. Hover your mouse over a node to see detailed information.

User question → Identify Claude's intent → RAG search → Vector DB similarity search → Evidence-based final response

AI Core Tool / API RAG / Vector Cloud / Service inactive active flow data packet
AI Project Delivery Framework

Project delivery

We bring AI agents, RAG, LLMOps, DevOps, and MLOps together to deliver enterprise AX and DX transformation.

AX / DX / AI Automation

Four-step execution framework from design to operation

AX DX DevOps MLOps

01 Discover

Business/data diagnosis

Analyze current processes, documents, databases, APIs, and infrastructure to define AI application scope and priorities.

As-Is analysisdata mapKPI alignment

02 Design

AI Agent · RAG Design

Create a reliable response system by designing prompt strategies, tool calls, permission flows, vector databases, and search structures.

PromptRAGGuardrail

03 Deploy

Cloud implementation & deployment

We connect AWS/Azure infrastructure, CI/CD, APIs, access control, and audit trails for enterprise deployment.

AWSCI/CDSecurity

04 Optimize

Operation Optimization

We monitor quality, cost, response speed, and failure indicators and continuously improve from a DevOps, MLOps, and AIOps perspective.

ObservabilityCostFeedback Loop

key deliverables

AI PoC, RAG knowledge base, Agent workflow, operation dashboard, cloud deployment automation

Area of ​​application

Learning platforms, enterprise workflow automation, dashboards, cloud migration, and AI services

Operating standards

Practical delivery that considers security, traceability, cost, scalability, and adoption

Project history (2001–2026)

25 years of experience in AI, cloud, education, public, and corporate projects

Total 90+ projects 2001–2026

technology stack

We design proven automation architectures spanning Claude (Anthropic), AI technologies, cloud platforms, and DevOps.

FINTECH STACK

Fintech project stack

Technologies shown in the public repositories, from financial data ingestion through analytics, retrieval, and deployment.

Financial data

KRX · pykrx · yfinance · OpenDART · pandas · NumPy

Quant & forecasting

QuantConnect LEAN · scikit-learn · LightGBM · MLP · portfolio optimization

AI & retrieval

FastAPI · Qdrant · pgvector · Ollama · ReAct agents

Apps & operations

PostgreSQL · Redis · Vanilla JS · ApexCharts · Docker Compose · AWS

Core AI

LLM · Foundation Models

Claude Opus 4 / Sonnet 4 / Haiku Claude Code (Anthropic) MCP (Model Context Protocol) GPT-4o / GPT-4-turbo Gemini 1.5 Pro HuggingFace OSS LLM Ollama · vLLM (on premises) Amazon Bedrock Azure OpenAI Service

AI Agent & Orchestration

LangChain / LangGraph LlamaIndex AutoGen · CrewAI n8n · Dify Cursor · GitHub Copilot Amazon Q Developer

RAG & Vector DB

Pinecone Weaviate Qdrant pgvector (PostgreSQL) ChromaDB · FAISS OpenSearch kNN Cohere Rerank

AI Coding & Development Automation

Claude Code Cursor IDE MCP Server development GitHub Copilot Automatic code review Automatically generate test code

MLOps & LLMOps

MLflow LangSmith Weights & Biases Amazon SageMaker Kubeflow Helicone · Arize Phoenix Prometheus · Grafana

Cloud & Data Platform

AWS Lambda · Step Functions S3 · API GW · ECS/EKS Glue · Athena · Redshift Azure DevOps · AKS Kafka · Kinesis dbt · Airflow (MWAA)

Agentic enterprise RAG system

Internal knowledge integration Agent-led retrieval and reasoning Permission-aware search and citations Business system and tool integration On-premises LLM operations Audit logs and quality monitoring
Core technologies Supporting technologies
Contact

Project Inquiry

Feel free to email us about AI agents, RAG, cloud infrastructure, or DX/AX transformation. We will quickly review and contact you.

info@edumgt.co.kr

Room 1206, Building E, Hyundai Terrace Tower, Yeonmujang 5-ga-gil, Seongdong-gu, Seoul