I started programming professionally in 2005. Back then, we were "Webmasters" - responsible for everything: HTML, CSS, backend, databases, deployment, sometimes even design.
As the industry evolved, we specialized. Frontend Devs, Backend Devs, DevOps, DBAs, Product Owners, UX Designers emerged... A complete project started requiring entire squads.
And now? AI is bringing the Webmaster back.
The Project
Over the past few months, I've shipped a project that had been shelved for years: a complete fitness platform. Solo. Taking on every role.
This isn't a half-baked MVP. It's a real product, published on the stores, with active users.
The Tech Stack
π± Mobile App (iOS + Android)
{
"react-native": "0.81.5",
"expo": "~54.0.31",
"typescript": "~5.9.2",
"@tanstack/react-query": "^5.85.9",
"i18next": "^25.8.0",
"expo-notifications": "~0.32.16",
"expo-local-authentication": "~17.0.8",
"react-native-maps": "1.20.1",
"react-native-purchases": "^9.7.1"
}
Key features:
- Push notifications with deep linking
- Biometric authentication (Face ID / Fingerprint)
- Maps with gym geolocation
- Monetization via RevenueCat
- Multi-language support (i18n)
- Camera with social post frames
βοΈ Backend API
# requirements/base.txt
Django==5.2.5
djangorestframework==3.16.1
channels==4.0.0 # Real-time WebSocket
daphne==4.0.0 # ASGI server
celery # Background tasks
redis # Cache + message broker
google-genai # Gemini AI
mediapipe # Body pose detection
stripe # Payments
Architecture:
- REST API with Django REST Framework
- WebSocket for real-time chat and notifications
- Celery + Redis for async processing
- JWT authentication
- PostgreSQL in production
π₯οΈ Admin Web Panel
{
"react": "^18.3.1",
"vite": "^6.0.5",
"@tanstack/react-query": "^5.62.0",
"@tanstack/react-table": "^8.20.0",
"tailwindcss": "^3.4.17",
"shadcn/ui": "latest",
"recharts": "^2.14.0",
"react-hook-form": "^7.54.0",
"zod": "^3.23.0"
}
βοΈ Infrastructure
- Hosting: Google Cloud Run (containerized)
- Storage: GCS (Cloud Run) + Backblaze B2 (production)
- CI/CD: EAS Build for mobile
- Monitoring: Native GCP logs
47+ Features Implemented
Some of the most interesting ones:
| Feature | Technology |
|---|---|
| AI body composition analysis | MediaPipe + Gemini |
| Gamified rankings | Points system with expiration |
| Instagram-style stories | React Native + Backend |
| AI-generated workouts | Google Gemini |
| Real-time chat | Django Channels + WebSocket |
| Push notifications | Expo + Celery |
| Biometric login | expo-local-authentication |
The Dilemma: Another Dev vs AI
Here's the reflection that intrigues me the most:
When I need to implement something new, I find myself thinking:
"Should I write the spec and hand it off to another dev, or specify it directly for AI?"
The answer has been consistently: specifying for AI is faster.
Not because AI is perfect - it makes mistakes, quite a few actually. But the feedback loop is instant.
With another dev:
Specify β Align β Wait β Review β Adjustments β Deploy
Time: days to weeks
With AI:
Specify β Generate β Review β Correct β Iterate β Deploy
Time: hours
The critical difference: the feedback loop is immediate.
There's no:
- Alignment meetings
- Timezone differences
- "Creative interpretations" of requirements
- "I'll pick it up next sprint"
The New "Webmaster"
In a way, we're back to the beginning. But with superpowers.
Today I work as:
| Role | What I do |
|---|---|
| Developer | Mobile + Backend + Web |
| DevOps | CI/CD, Cloud, Containers |
| DBA | Modeling, Migrations, Queries |
| Product Owner | Specs, Prioritization, Roadmap |
| QA | Testing, Code Review |
The difference is I have an "infinite intern" who:
- Never complains about repetitive tasks
- Is available 24/7
- Improves every month
- Doesn't need onboarding
Lessons Learned
For experienced devs
AI won't replace those who understand:
- System architecture
- Technical trade-offs
- How to ask the right questions
- When AI is wrong
It amplifies your capabilities. The more you know, the more you can extract from it.
For beginner devs
Learning the fundamentals has never been more important.
AI is a powerful lever - but you need to know where to point it. Without solid fundamentals, you can't:
- Validate generated code
- Identify subtle bugs
- Ask the right questions
- Debug when it breaks
Conclusion
21 years after starting, I'm building more ambitious projects, faster, and having more fun.
The Webmaster is back. Only now, he has AI.
What about you?
Have you been using AI in your development workflow? What has been your experience?
Drop a comment - I want to hear stories from those experimenting with this new paradigm.
If you enjoyed this, follow for more content on solo development with AI, React Native, and Django.
Top comments (0)