| Production SaaS |
99% Uptime |
$480/month MRR |
31 Paying Subscribers |
| Built: 2024 |
Timeline: 3 months |
Status: Archived (ran profitably for 220 days) |
📊 Overview
Anubis Bot was a 71-microservice cryptocurrency intelligence platform that served 31 paying subscribers with real-time market data, ML predictions, and automated trading signals via Telegram.
The Achievement:
- Built and operated a production SaaS generating $480/month recurring revenue
- Maintained 99% uptime over 220 consecutive days
- Processed 54,000+ events daily with zero data loss
- Managed complex multi-service architecture as solo developer
🎯 The Problem
Cryptocurrency traders needed:
- Real-time data across multiple exchanges and chains
- ML predictions for market movements
- Automated alerts for trading opportunities
- Mobile access during market volatility
- Reliable service (99%+ uptime, no missed signals)
Existing solutions were either:
- Too expensive ($50-200/month)
- Too complex (required technical expertise)
- Unreliable (frequent downtime)
- Desktop-only (not mobile-friendly)
💡 The Solution
71-microservice architecture orchestrating:
- Market data aggregation (multiple exchanges)
- ML ensemble models (price predictions)
- Risk analysis (portfolio optimization)
- Telegram delivery (6 different bots)
- Stripe payments (subscription management)
- Database management (29GB PostgreSQL)
Pricing: $15/month (3-10x cheaper than competitors)
Interface: Mobile-first Telegram bots
Reliability: 99% uptime, zero data loss guarantee
🏗️ Technical Architecture
Scale
- 71 microservices in production simultaneously
- 54,000+ events processed daily
- 29GB PostgreSQL database (1.86M rows, 309 columns)
- 6 Telegram bots (different subscriber tiers)
- 31 paying subscribers at peak
- 394,767 lines of Python code written
Reliability
- 99% uptime over 220 consecutive days
- Zero data loss across all deployments
- Real-time WebSocket processing
- Sub-3-minute response times maintained
- Automated health monitoring
- Comprehensive backup systems
Business Operations
- $480/month MRR from paying customers
- Stripe integration (payment processing)
- Subscription management (tier-based access)
- Customer support via Telegram
- Real transactions (not demo/test)
🛠️ Technology Stack
Backend:
- Python 3.x (asyncio for concurrent processing)
- PostgreSQL (29GB production database)
- Redis (caching and pub/sub)
- Docker (containerization)
- SystemD (service management)
APIs & Integrations:
- CoinMarketCap API
- CoinGecko API
- Multiple exchange WebSockets
- Stripe (payments)
- Telegram Bot API (6 bots)
- Google Drive (data exports)
ML/AI:
- Ensemble prediction models
- Time-series analysis
- Portfolio optimization
- Risk scoring
Infrastructure:
- DigitalOcean VPS
- Automated deployment
- Health monitoring
- Backup systems
💼 Key Technical Challenges
Challenge 1: Real-Time Data at Scale
Problem: 54,000 events/day across multiple sources
Solution: Async Python with WebSocket connections, event queue, batched processing
Result: Sub-3-minute latency maintained, zero dropped events
Challenge 2: Database Optimization
Problem: 29GB database, 1.86M rows, complex queries
Solution: Autovacuum tuning, dead tuple management, index optimization
Result: Query times <100ms, database stable at scale
Challenge 3: 99% Uptime Requirement
Problem: Paying customers expect reliability
Solution: Automated health checks, graceful degradation, comprehensive error handling
Result: 99% uptime over 220 days, zero data loss
Challenge 4: Multi-Service Orchestration
Problem: 71 microservices need coordination
Solution: Event-driven architecture, service discovery, centralized logging
Result: Services run independently, failures contained
📈 Business Results
Revenue:
- $480/month MRR at peak
- 31 paying subscribers
- $15/month base tier
- Profitable operation (revenue > costs)
Operations:
- 220 consecutive days in production
- Customer support via Telegram
- Feature requests implemented
- Subscription management automated
Why I Shut It Down:
- Time commitment (10+ hours/week maintenance)
- Market shift (crypto bear market reduced demand)
- Opportunity cost (could earn more as freelancer)
- Learning complete (achieved SaaS goal)
Key Lesson: Built a real business, learned operations, proved I can ship production systems.
🎓 What I Learned
Technical:
- PostgreSQL optimization at scale
- Async Python mastery (asyncio, concurrent processing)
- Multi-service architecture patterns
- Real-time data processing
- Production database management
Business:
- Customer acquisition (marketing, positioning)
- Subscription pricing (what customers will pay)
- Customer support (response times, issue resolution)
- Feature prioritization (what actually matters)
- Churn management (why customers leave)
Operations:
- 24/7 monitoring (when to wake up vs when to wait)
- Backup strategies (what to back up, how often)
- Deployment automation (zero-downtime updates)
- Incident response (how to handle failures)
- Cost management (AWS vs DigitalOcean tradeoffs)
🏆 Why This Matters
This wasn’t a side project or demo.
This was a real business with:
- Real customers paying real money
- Real uptime requirements (99%+)
- Real data (no mocks or fixtures)
- Real transactions (Stripe payments)
- Real support burden (customer questions)
Proves I can:
- Build production systems that run 24/7
- Handle real transactions and payments
- Maintain 99%+ uptime
- Manage databases at scale
- Ship features customers actually use
- Operate a business, not just write code
📊 Metrics Summary
| Metric |
Value |
| Uptime |
99% over 220 days |
| Revenue |
$480/month MRR |
| Customers |
31 paying subscribers |
| Events |
54,000/day |
| Database |
29GB, 1.86M rows |
| Services |
71 microservices |
| Telegram Bots |
6 bots |
| Data Loss |
Zero |
| Lines of Code |
394,767 (Python) |
- Praxis Bot - Solana DeFi monitoring (99.9% accuracy)
- FreelanceOS - Upwork automation (80% time reduction)