Moe Loubani
Software Developer
Development Consultant
- Phone: 1-877-264-0025
- Email: moe@moe.ca
- Location: Waterloo, Ontario, Canada
RAG, Agents, MCP Tools, Voice AI, and more
My Portfolio
Websites and services I've created, own, and manage
Cantal
Canadian Tech Talent MarketplaceCantal is a comprehensive talent marketplace that bridges the gap between US organizations and exceptional Canadian tech professionals. The platform streamlines the entire recruitment process, from initial candidate discovery to successful placement, while handling compliance, payroll, and ongoing relationship management. Built with scalability and user experience in mind, Cantal serves both early-stage startups and Fortune 500 companies.
Key Features
- Connecting US organizations with exceptional Canadian talent across AI, full-stack development, and DevOps roles
- AI Consulting, Augmented Staffing, and Digital Transformation services with proven ROI metrics
- Tech stack: Next.js 14, Node.js, PostgreSQL with Prisma ORM, Stripe for payments and subscriptions
- Custom talent matching algorithm using vector embeddings and semantic search for precise candidate-role alignment
- Automated deployment workflows with Railway, continuous integration via GitHub Actions
- Real-time chat system for client-talent communication, document management, and interview scheduling
Technical Challenges
The main challenge was creating a matching system that could accurately pair candidates with roles based not just on technical skills, but also on cultural fit, timezone compatibility, and project requirements. Additionally, handling international payments and tax compliance required careful integration with multiple financial services.
Solution Approach
Implemented a multi-factor scoring system that weighs technical skills, experience level, industry background, and soft skills. Built a comprehensive admin dashboard for managing the entire talent pipeline, from application to placement and ongoing performance tracking.
Naami
Personalized Story Videos for ChildrenNaami transforms parenting challenges into engaging storytelling experiences. Parents upload their child's photo and describe a behavior they want to address or an achievement they want to celebrate. Naami's AI creates personalized 6-panel storyboards featuring the child as the hero, making learning and positive reinforcement both effective and memorable. The platform has helped thousands of parents address bedtime routines, sharing, listening, and other common childhood challenges.
Key Features
- AI-powered platform creating custom 6-panel storyboards featuring children as the heroes of their own stories
- Helps parents address behavioral challenges (bedtime, sharing, listening) and celebrate achievements through personalized narratives
- Tech stack: Next.js 14 with App Router, Prisma for database management, PostgreSQL for data persistence, Redis for caching and queue management
- Replicate AI for image generation and style transfer, creating consistent character representations across panels
- ElevenLabs voice synthesis for narrated story versions, Stripe for subscription payments, NextAuth for secure authentication
- Background job processing for story generation, email delivery system for completed stories
- Responsive design optimized for mobile parents on-the-go
Technical Challenges
Generating consistent character representations across multiple story panels while maintaining the child's likeness was technically complex. Additionally, ensuring stories were age-appropriate, positive, and effective required careful prompt engineering and content moderation.
Solution Approach
Developed a multi-stage generation pipeline that first creates character references, then generates individual panels with style consistency prompts. Implemented a content review system with automated checks for age-appropriate language and positive messaging. Added parent feedback loops to continuously improve story quality and effectiveness.
ReceptionIQ
AI-Powered Virtual ReceptionistReceptionIQ eliminates missed calls and lost revenue for service-based businesses by providing an AI receptionist that never sleeps, never takes breaks, and handles every call professionally. The system answers calls, qualifies leads, books appointments directly into business calendars, and routes urgent matters to the right team members. Businesses using ReceptionIQ report capturing 87% more leads and saving over 200 hours monthly on phone management.
Key Features
- 24/7 AI receptionist answering calls, qualifying leads, and booking appointments automatically
- Natural conversation powered by OpenAI's GPT-4 with custom fine-tuning for business-specific responses
- Tech stack: Next.js for the dashboard, Twilio for voice communication and SMS, WebSockets for real-time updates
- Calendar integrations with Google Calendar, Outlook, and iCal for seamless appointment booking
- Real-time call routing based on caller intent, business hours, and team availability
- Comprehensive analytics dashboard showing call volume, conversion rates, peak times, and revenue captured
- CRM integration with popular platforms for automatic lead capture and follow-up workflows
- Multi-location support with location-specific greetings, services, and staff routing
Technical Challenges
Creating natural-sounding conversations that could handle diverse caller intents, accents, and speaking styles while maintaining professionalism was crucial. The system needed to gracefully handle interruptions, background noise, and unclear requests without frustrating callers.
Solution Approach
Implemented a streaming speech-to-text pipeline with real-time processing and context maintenance. Built a conversation state machine that tracks dialogue flow and can recover from interruptions. Added voice activity detection to handle natural pauses and overlapping speech. Created a comprehensive testing framework with diverse caller scenarios to continuously improve conversation quality.
Protect Uploads
Enterprise WordPress Media Security PluginProtect Uploads Pro is an enterprise-grade WordPress security plugin that transforms media libraries into fortresses. Building on a popular free plugin with 30,000+ downloads, the Pro version adds advanced protection layers, business intelligence dashboards, and professional-grade security features. The platform serves photographers, agencies, e-commerce sites, and enterprises who need to protect valuable digital assets while maintaining user experience and SEO performance.
Key Features
- 10 layers of protection including right-click prevention, drag-and-drop blocking, DevTools protection, and source code obfuscation
- Military-grade password protection with 4 password types: standard, single-use, limited-use, and time-expiring passwords
- Business intelligence dashboard with download analytics, geographic access patterns, security incident tracking, and ROI metrics
- Advanced watermarking engine with text and logo support, 9 position options, opacity control, and batch processing capabilities
- Tech stack: WordPress/PHP architecture, Stripe for licensing and subscriptions, GD Library for image processing
- Smart hotlink detection with referrer analysis, whitelist/blacklist management, and custom response options
- Multisite-ready architecture supporting enterprise networks with centralized control and site-specific overrides
- Role-based access control with WordPress integration, custom capabilities, and conditional logic rules
Technical Challenges
Creating a security solution that was both powerful enough for enterprise use and simple enough for small businesses was challenging. The plugin needed to protect media files without breaking legitimate use cases, CDN compatibility, or SEO. Additionally, the watermarking system had to process images efficiently without degrading quality or slowing down websites, even on shared hosting environments.
Solution Approach
Developed a modular architecture that allows users to enable only the protection features they need. Implemented server-side processing with intelligent caching to minimize performance impact. Created a comprehensive compatibility layer that works with major page builders, caching plugins, and CDN providers. Built a smart detection system that differentiates between legitimate access and theft attempts, ensuring SEO-friendly operation while blocking hotlinking.
Instant Expert
AI Chatbot with Automatic Lead CaptureInstant Expert is a SaaS platform that adds AI-powered chatbots to websites, answering visitor questions using only the website's own content while naturally capturing lead information. Powered by Claude AI and advanced RAG technology, the system automatically scrapes and indexes website content, then provides accurate, contextual answers. Unlike generic chatbots, Instant Expert ensures responses are based exclusively on the client's content, maintaining brand consistency and accuracy while converting visitors into qualified leads.
Key Features
- Claude AI-powered conversation engine providing accurate answers based exclusively on website content
- Advanced RAG (Retrieval-Augmented Generation) technology with automatic website scraping and intelligent content indexing
- Natural lead capture during conversations without requiring traditional forms or interrupting user experience
- Tech stack: Next.js 14 for web application, Python for RAG pipeline, Anthropic Claude API, vector database for semantic search
- One-line widget integration requiring no technical expertise, with full customization options for colors, prompts, and branding
- Comprehensive lead management dashboard with conversation history, export capabilities, and analytics
- Multi-website support with tiered pricing from free (100 conversations) to enterprise (unlimited)
- Custom branding options including white-label solutions for enterprise clients
Technical Challenges
Building a RAG system that could accurately extract and index content from diverse website structures while maintaining context was complex. The chatbot needed to naturally capture lead information without feeling pushy or breaking conversation flow. Additionally, the system had to handle websites with dynamic content, JavaScript-rendered pages, and various anti-scraping measures while respecting robots.txt and rate limits.
Solution Approach
Implemented a sophisticated web scraping pipeline using Playwright for JavaScript-rendered content and BeautifulSoup for static pages. Developed a chunking strategy that preserves context while optimizing for semantic search. Created a conversation flow that uses Claude's natural language capabilities to identify moments when lead capture feels organic. Built a robust indexing system that can handle content updates and supports manual content management for clients with complex websites.