☰ Network & Cloud

Network & Cloud Projects

SD-WAN Migration

MPLS to SD-WAN network transformation for a freight forwarder with multiple remote sites.

Project Profile

A global freight forwarding company operating 20 branch offices in China, experienced increasing network congestion and high operational costs due to its centralized MPLS architecture. All remote offices were required to route traffic through the Shanghai data center, creating significant bandwidth bottlenecks and limiting the deployment of new collaboration systems such as Cisco Webex video conferencing. The objective of this project was to modernize the network architecture by migrating from a fully centralized MPLS topology to a hybrid SD-WAN architecture, improving performance, reducing cost, and enabling local internet access for branch offices.
Original MPLS

Customer Pain Points

MPLS Bandwidth Bottleneck

High MPLS Cost

Lack of Local Internet Access due to centralized security controls


SD-WAN

Solution

Deploy FortiGate firewall at each remote office

Maintain MPLS for critical traffic

Use IPsec VPN over internet for non-critical traffic

Enable local internet breakout

IP-Transit

IP Transit Connectivity Enhancement Project

Project Profile

The customer is a leading mobile network operator in Hong Kong, serving approximately 3.5 million mobile subscribers across the territory. Due to rapid growth in mobile data traffic and increasing cross-border internet usage, the operator experienced capacity bottlenecks and latency issues in its existing connectivity with China networks and international upstream carriers. To improve network resilience, routing flexibility, and international connectivity performance, the operator initiated an IP Transit upgrade project.
Original MPLS

Project Objective

Increase international internet capacity

Eliminate congestion between Hong Kong and overseas networks

Improve routing efficiency and latency for China and international traffic

Enhance network resilience through redundant transit paths


Solution Overview

Deploy dual-path 10Gbps IP Transit connectivity to ensure high availability.

IP Prefix advertisement to upstream carriers

Route optimization for international and China-bound traffic

Cloud

Hybrid Cloud Solution

Project Profile

Customer is a leading provider of integrated logistics solutions for the fashion industry in China. With the rapid growth of fashion e-commerce and retail logistics operations, the company requires high availability, reliable data protection, and scalable IT infrastructure to support its warehouse management and logistics systems.
HybirdCloud

Challenges

The current Oracle license cannot be migrated directly to the public cloud.

All systems must comply with the company's "3+2+1" backup policy.

Improve routing efficiency and latency for China and international traffic

24×7 availability requirement

Project Objective

Hybrid Cloud Infrastructure

Virtualized compute environment using VMware ESXi

Data protection using Veeam Backup & Replication

24×7 monitoring and support services with 3+2+1 backup strategy


Supply Chain & EDI

EDI Migration

Project Profile

Implement EDI integration with a new retail partner to enable automated and reliable electronic document exchange between the retailer, supplier ERP system, and warehouse management system (WMS). The integration will support the end-to-end order lifecycle including purchase order processing, inventory visibility, warehouse fulfillment, shipment notification, and invoicing, using standardized EDI transactions.

Electronic Data Interchange (EDI) is a critical component for modern trading and e-commerce operations. It allows trading partners to exchange business documents electronically in standardized formats, eliminating manual data entry and improving transaction speed, accuracy, and operational efficiency.

For high-volume retail and e-commerce channels, EDI supports automated order processing, real-time inventory visibility, faster warehouse fulfillment, accurate shipment notifications, and automated invoice transmission. In modern omnichannel retail environments, EDI provides scalable and standardized system-to-system communication that helps reduce manual errors, improve service levels, and meet retailer compliance requirements.
HybirdCloud

Data Flow Overview

Retailer → Supplier → Warehouse → Supplier → Retailer

  1. Supplier (We) sends 846 Inventory Advice as required
  2. Retailer sends 850 Purchase Order
  3. Supplier (We) sends 855 Purchase Order Acknowledgment
  4. Supplier (We) converts order details into 940 Warehouse Shipping Order
  5. Warehouse processes picking, labeling, and shipping
  6. Warehouse sends 945 Warehouse Shipping Advice after the product is shipped.
  7. Supplier (We) generates and sends 856 Advance Ship Notice (ASN) with tracking information.
  8. Supplier (We) sends 810 Invoice to Retailer

Implementation Activities

Planning

  • Define project requirements with retailer and internal stakeholders
  • Confirm document standards, mandatory fields, packing standards , label requirements and business rules
  • Define integration architecture between EDI platform, ERP, and WMS

Mapping & Development

  • Configure EDI mappings for 850, 860, 997, 846, 855, 856, 810, 940, and 945
  • Develop integration between EDI platform and retailer system (AS2 or SFTP)
  • Develop integration between supplier ERP and warehouse management system
  • Configure packing list and shipping label templates based on retailer requirements

Testing

  • EDI Connectivity Testing
  • Document Mapping Validation
  • Unit Testing
  • Warehouse Integration Testing
  • End-to-End Order Flow Testing / UAT (User Acceptance Testing)

Deployment

  • Prepare production cutover plan
  • Execute go-live migration to production environment
  • Monitor initial transactions and resolve post-go-live issues

Major Retailers

Amazon
Walmart
QVC
Lowe's
Kohl's
JCPenney
Dollar Tree
Home Depot
Target
Academy Sports
Wayfair
Meijer
True Value
Costco

AI Chat Bot with LangChain & Gradio

Use LangChain & OpenAI LLM module to call OpenAI and display response.


Tools: Gradio + OpenAI API

Project Profile

This project is a lightweight AI chatbot application built using LangChain, Gradio, and the OpenAI API. It provides an interactive web-based chat interface where users can input prompts and receive intelligent responses powered by a Large Language Model (LLM).
please click here to access this project in my github

langchain

Interactive chatbot UI using Gradio

LLM integration via OpenAI API

Modular pipeline using LangChain

Prompt templating for structured input



Chat with PDF Application (AI & RAG)

Project Profile

The application uses a Retrieval-Augmented Generation (RAG) architecture to ensure the LLM answers are based only on the uploaded PDF document.

This project is a PDF Chat Application that allows users to:

  • Upload a PDF document
  • Ask questions in natural language
  • Receive answers generated by an AI model
  • View source page references

Tools: LangChain + Gradio + OpenAI API


please click here to access this project in my github

chatWithPDF

How it Works

load PDF using PyPDFLoader

Split text into chunks

Convert chunks into vector format & store in vector database (Chroma)

Question and Answering (RAG processing)

Retrieve relevant information from Chroma

Use Prompt Template with contect and question to build up Prompt

Generate answer using LLM



Powered by Ivan Chiu