Verified evidence01
Full-stack application architecture
The repository documents the Vue, FastAPI, database, container, and MT5 integration boundaries
Source · Public GitHub repository
Vue 3FastAPIDockerMariaDB
Project archive
Project case study · Full-Stack
Trading SaaS Platform
Multi-tenant trading SaaS connecting a Vue interface and FastAPI services on Ubuntu to an MT5 execution environment on Windows through an encrypted site-to-site VPN.
Vue 3QuasarTypeScriptFastAPIPythonMetaTrader 5MariaDBWebSocketsDockerSite-to-Site VPNUbuntuWindows Server
2024
Read this case as a
Choose your lens
Outcome-focused delivery
Owned work across frontend, Python services, database integration, container deployment, and cross-operating-system infrastructure for a technically complex trading platform.
Evidence mode · on
Claims come with context
Each outcome shows its source, related capability, and confidence level. Missing operational data stays visibly unavailable.
Supporting context02
Cross-platform execution connectivity
An encrypted site-to-site VPN connects Ubuntu application services with the Windows-hosted MT5 environment
Source · Infrastructure addresses and credentials excluded
VPNUbuntuWindows Server
Supporting context03
Live portfolio state
WebSocket channels deliver changing market, account, and portfolio information to the client interface
Source · Private operating environment
WebSocketsVue 3Financial systems
Skill → proof
Where each capability shows up
Vue 3Full-stack application architecture
FastAPIFull-stack application architecture
DockerFull-stack application architecture
MariaDBFull-stack application architecture
VPNCross-platform execution connectivity
UbuntuCross-platform execution connectivity
Windows ServerCross-platform execution connectivity
WebSocketsLive portfolio state
Vue 3Live portfolio state
Financial systemsLive portfolio state
01 / Problem
A fragmented workflow slowed the whole system.
The product needed to connect modern web services running on Ubuntu with MetaTrader 5 processes that required Windows Server. The connection had to remain secure and responsive while transferring market, account, and execution state between different operating environments.
02 / Solution
One product model, expressed clearly at every layer.
I engineered a full-stack platform using Vue, Quasar, FastAPI, Python, MariaDB, and WebSockets. A site-to-site VPN connects the Ubuntu application environment to the Windows-hosted MT5 execution engine, while separated service boundaries handle portfolio data, risk controls, and execution communication.
Architecture replay
Walk through the system
Select a layer to inspect its responsibility.
Active layer 1
Vue 3 and Quasar trading interface
Vue 3 and Quasar trading interface owns the user-facing workflow and consumes a stable contract.
Responsibility stays explicit at this boundary.
Project media story
The system, seen in use
Artifacts are captioned with the engineering story they support.
Trading dashboard
Portfolio and market information presented through the main SaaS interface using demonstration data.
Risk-control workflow
Interface for reviewing trading limits, execution state, and risk information.
Cross-platform architecture
Ubuntu services communicate securely with the Windows-hosted MT5 execution environment.
Decision log
Isolate the MT5 execution environment
- Context
- MT5 required Windows while the application services ran on Ubuntu
- Decision
- Connected the environments through an encrypted site-to-site VPN
- Trade-off
- Added networking and monitoring complexity but preserved platform-specific execution requirements
Decision log
Separate execution from application services
- Context
- Trading execution failures should not directly destabilize the web application
- Decision
- Kept execution, API, data, and interface responsibilities in separate boundaries
- Trade-off
- Required explicit synchronization but improved system isolation
03 / Challenges overcome
Complexity moved behind durable boundaries.
The most difficult work involved cross-platform routing, firewall configuration, VPN stability, and recovery behavior when connections changed. The platform also needed to prevent infrastructure failures from producing misleading trading state or unsafe execution behavior.
Next step
Want to discuss a similar system?
Share the context and we’ll find a useful first move.