The Platform
Exlink is an enterprise expert network platform serving knowledge industry businesses, managing expert networks, project workflows, scheduling, billing, compliance, and communication with deep integrations into Zoom, Microsoft Teams, Google Workspace, and OpenAI. The platform runs on Laravel with a Vue 3 frontend, PostgreSQL with pgvector, and schema-based multi-tenancy isolating each client’s data at the database level.
This is one of the most complex platforms I’ve worked on: 130+ Eloquent models, 340+ migrations, 40+ service classes, and 21 background jobs serving tens of thousands of active users across expert networks and client research teams. I joined as one of two developers and grew into the team lead role, responsible for architecture decisions, backend development, infrastructure, DevOps, and leading distributed engineering teams. I was a key decision maker in hiring, onboarding new developers, and shaping the technical direction of the platform.
Architecture
The backend combines a Laravel application with DDD-structured domains alongside legacy PHP code (both actively serving the platform. The newer domain layer handles AI, time tracking, file parsing, transcripts, and project chat, while the established codebase covers core expert management, project workflows, and administrative operations. The Vue 3 frontend is a standalone SPA with TypeScript, Pinia state management, and real-time updates via Pusher and Laravel Echo.
Core technical decisions
- Multi-tenant isolation: schema-based PostgreSQL separation with dynamic connection switching per request via ClientConnectionService
- DDD domains: Expert, FileParsingAI, ProjectAIChat, TimeTracking, Transcripts, Cloudflare, and InexOne as bounded contexts
- PostgreSQL with pgvector: relational data with vector embeddings for AI-powered expert matching
- Meilisearch: full-text search across expert profiles via Laravel Scout
- Vue 3 + TypeScript: Vite SPA with Reka UI, Vee-Validate + Zod, vue-i18n, and TipTap rich text
- Real-time: Pusher WebSockets with Laravel Echo for live notifications and updates
- Docker: containerized with Nginx, PHP-FPM, PostgreSQL 17, and Meilisearch
What I Built
Expert Network Management
- Expert registration, profiles, and badge/certification management with custom fields
- AI-powered CV parsing and document extraction for expert onboarding
- Expert search with Meilisearch full-text indexing
- Availability scheduling and vetting questionnaire system
- OpenAI-powered expert suggestions for project matching
Project & Workflow Engine
- Project creation with expert selection, briefs, and custom templates
- AI project chat for consultation conversations
- KPI tracking and project reporting
- Compliance management with GDPR consent tracking and blacklisted company handling
Communication & Scheduling
- Zoom: meeting scheduling, recordings, transcripts with automated delivery
- Microsoft Teams: meetings, recordings, and webhook processing
- Twilio: voice call integration with call tracking
- Google Calendar & Outlook: bi-directional calendar sync via Cronofy
- Post-call feedback forms and call reminder automation
Billing & Financial Management
- Billing records with approval workflow and invoice generation
- Multi-currency support with exchange rate management
- Expert rate management and package deals
- Subscription management across clients
Platform Infrastructure
- Schema-based multi-tenancy with operations across all tenant schemas
- SAML SSO with Google and Microsoft, plus OAuth and 2FA
- Role-based access control with subroles and delegated permissions
- Google BigQuery integration for centralized logging and analytics
- SendGrid email delivery with failed email retry queue
- Data export system for Excel and PDF reporting
Security & Compliance
- Regular security audits across the platform
- GDPR consent tracking and data handling compliance
- Blacklisted company enforcement and data residency controls
Infrastructure & DevOps
Beyond the application, I designed and manage the entire infrastructure: 15+ servers across multiple providers and geographies, including client-specific deployments with strict data residency requirements and isolated development environments where source code never leaves controlled infrastructure. CI/CD pipelines automate deployment for the landing page and white-labeled Zoom SDK, while the core platform uses on-demand deployment with setup scripts due to infrastructure constraints.
Results
- Enterprise platform serving tens of thousands of active users with full data isolation
- Grew from 2-person team to team lead managing distributed engineers across multiple countries
- 130+ models across expert management, projects, billing, compliance, and communication
- Deep integrations with Zoom, Teams, Google Workspace, and Microsoft 365
- AI-powered expert matching and document parsing reducing manual onboarding time
- 15+ servers across multiple providers and geographies, managed end-to-end
- Regular security audits maintaining platform integrity