Real Projects

This module is where everything comes together. I am going to walk you through my most significant projects — the ones that taught me the most and the ones that are running in production right now. These are not toy demos. These are real applications solving real problems for real people.

SMTPCloud Dashboard is the flagship. It is a full B2B SaaS platform for email infrastructure management. The numbers tell the story: 44,000+ lines of code, 145 git commits, 30+ database tables. It is built with Next.js 16 on the frontend, Express on the backend, and PostgreSQL for the database. The frontend deploys to Vercel, the backend runs on a Hetzner server.

What does SMTPCloud Dashboard actually do? It lets businesses manage their email campaigns end to end. You create a campaign, upload or select your recipient list, configure sending parameters (which domain to send from, how fast to send, time zones), and launch. The system tracks everything in real time — opens, clicks, bounces, unsubscribes. It has A/B testing so you can compare subject lines or content variations. It has bot detection to filter out fake opens from security scanners. It integrates with Google Postmaster Tools to show you how Google sees your sending reputation. And it has smart email throttling by domain — so you do not blast 100,000 emails to Gmail all at once and get blocked.

Building SMTPCloud Dashboard taught me more than everything else combined. It is where I learned about complex state management, real-time data updates, background job processing, rate limiting, and building systems that need to be reliable because real money depends on them. Every module in this course — frontend, backend, databases, infrastructure — I could illustrate with examples from this one project alone.

Quality Monitoring is an enterprise tool I built for INFUSE, the company I work at. The QA team needed a way to track and manage quality issues across campaigns. Before this tool, everything lived in Google Sheets. Quality Monitoring auto-syncs data from 7 different Google Sheets every 10 minutes, so the data is always current without anyone manually updating anything.

Quality Monitoring has AI-powered insights using the Claude API — it analyzes patterns in quality issues and generates actionable summaries. It sends automated daily email reports to team leads and managers. The whole thing has 111 automated tests to make sure nothing breaks when I add new features. It runs in Docker on AWS EC2 and is used by real teams every single day. This project taught me about enterprise requirements: role-based access control, audit trails, automated reporting, and building something that an entire team depends on.

MyVaultKeep.io is completely different from everything else I have built. It is a crypto trading bot with a web dashboard. The system connects to the Solana blockchain and executes trades on decentralized exchanges via the Jupiter API. It has a Telegram bot for quick commands and alerts, plus a full web dashboard for detailed portfolio management and trade history.

The technical architecture of MyVaultKeep.io is the most complex of any project I have built. It is a monorepo managed with Turborepo. It uses BullMQ for job queues (processing trades in the background), WebSocket connections for real-time price updates and trade notifications, and AES-256 encryption for storing private keys securely. There is even a referral system built in. Building a trading bot taught me about handling real money, the importance of error handling (a bug could lose funds), and working with blockchain APIs.

VaultKeep is a mobile app for collectors — people who collect cards, coins, memorabilia, that kind of thing. It is built with React Native and Expo, which means one codebase runs on both iOS and Android. The standout feature is AI item recognition using Claude Vision: you take a photo of a collectible and the AI identifies it, estimates its value, and suggests relevant details.

VaultKeep also has barcode scanning for quickly adding items, social features (a feed where collectors can share items, leave comments, and rate collections), and a Supabase backend for authentication and data storage. Monetization is handled through RevenueCat for managing subscriptions. Going from web development to mobile was a big jump, but React Native made it manageable because the core concepts (React components, state, props) are the same.

CampaignPulse is a QA automation tool that I am particularly proud of because of the time it saves. Before CampaignPulse, checking a campaign for issues — broken links, missing tracking parameters, DNS problems, content errors — took 30+ minutes of manual work per campaign. CampaignPulse does it in under 1 minute.

The interesting thing about CampaignPulse is that it is built with vanilla JavaScript. No React, no frameworks, no build tools. Just plain HTML, CSS, and JS. I built it this way intentionally to keep it simple and fast. It connects to n8n webhooks for workflow automation and uses GPT-4o for AI-powered content analysis. It is running in production at campaignpulse.infuse.com and the QA team uses it daily. Sometimes the simplest technical choice is the right one.

LeadTool 2.0 is my Python project. Built with FastAPI on the backend and AG Grid on the frontend for a powerful spreadsheet-like interface. It manages leads through 13+ different statuses (new, contacted, qualified, converted, and so on), has duplicate detection to prevent the same lead from being entered twice, and uses Celery with Redis for background task processing — things like bulk imports, data enrichment, and automated status updates.

LeadTool taught me that switching programming languages is not as hard as people think. The concepts are the same: you have routes that handle requests, a database that stores data, and a frontend that displays it. Python syntax is different from JavaScript, but Claude handles both equally well. If you can build with one language, you can build with any language — because the hard part was never the syntax. It was knowing what to build and how to describe it.

Looking across all these projects, the pattern is always the same. I had a problem or an idea. I described it to Claude. We built it piece by piece, starting with the core feature and adding complexity over time. Every project made me better at the next one — not because I memorized more syntax, but because I got better at thinking about architecture, anticipating problems, and communicating what I needed. That is the real skill this course is teaching you.