Presence-Aware Infrastructure: My Lab Knows When I'm Home (and Saves Me $$$)
Building the Ultimate R&D Cloud Lab: How I Achieved Real-Time Cost Control and Zero-Waste AWS Experimentation
TL;DR: Built a sophisticated AWS R&D environment with real-time cost tracking ($0.01/day for visibility), automated shutdown systems based on sleep detection, and seamless hybrid cloud integration. The result? Cost visibility + intelligent automation = zero waste and maximum experimentation velocity.
The R&D Cost Problem Every Technical Leader Knows 💸
Let's be honest traditional R&D labs suffer from two critical flaws that'll make your wallet weep:
No cost visibility until the monthly AWS bill lands like a financial haymaker
Always-on infrastructure burning cash 24/7, even when you're sleeping
I decided to solve this properly. No more monthly surprises, no more "oops, I left that GPU instance running for three weeks" moments.
The Solution: Real-Time Cost Intelligence That Actually Works
Daily Cost Tracking: The $0.01 Investment That Saves Thousands
Here's the game-changer: I know exactly what my experiments cost the day after I run them. Using AWS Cost Explorer API, my system delivers:
Immediate feedback loop on experimental costs
Daily spending attribution to specific tests and infrastructure changes
Proactive budget management instead of monthly financial surprises
The monitoring itself costs roughly $0.01 per day—an acceptable overhead that pays for itself within hours of preventing a single oversized experiment.
Why This Beats Every Alternative
Traditional Setup: Spin up instances → Run experiments → Hope for the best → Get shocked by the monthly bill
My Smart R&D Lab: Daily cost reports → Instant experiment attribution → Sleep-based automation → 60-80% cost savings
The difference? Real-time awareness creates behavioural change. When you see the daily cost impact of your experiments, you naturally optimize.
The Architecture: Hybrid Intelligence Done Right 🏗️
Multi-Location Home DC Foundation
My setup starts with a distributed home data centre (plss i really, really want to call them a DC even tho they are not) across 4 physical locations:
ARM hardware for lightweight processing
IoT and industrial systems for specialised testing
High-performance CPU/RAM for compute-intensive workloads
GPU hardware for GPU workloads
AWS Cloud Integration via WireGuard S2S VPN
The home infrastructure connects securely to AWS through WireGuard site-to-site VPN, enabling:
Secure data extraction from home lab to AWS for processing
Private workload testing without public exposure
Hybrid analytics process home-generated data using AWS services
VPC Architecture: Security Without Complexity
Public Subnet:
Bastion host (auto-managed based on presence)
NAT Gateway for private subnet access
Security groups locked to trusted IPs only
Private Subnet:
Development and testing instances
Database resources
No direct internet access—everything routes through NAT
The Automation That Changes Everything ⚡
Presence Detection: Your Lab Knows When You're Home
The system detects my presence through WiFi network monitoring when my phone connects to home WiFi, the automation triggers. This approach is:
Zero battery impact (uses existing WiFi connection)
Instant detection (phone connects within seconds of arriving)
100% reliable (can't fake being home unless you're actually there)
Sleep Detection: Smart Resource Management
Here's where it gets sophisticated. The system monitors network activity patterns to detect sleep cycles:
Network activity monitoring of phone and laptop usage
Pattern recognition when activity drops below the working baseline
Time-based logic (only between 20:00-04:00)
Multi-device correlation to avoid false positives
it uses my phone and laptop in an OR condition
if any one of the devices (laptop or phone ) is active means I am awake.
This works in a 30-minute internal to decide if I am awake
if any of my server in any of these locations are talking to the server in aws (in the 2 VPC) it keeps the AWS env active since i might be running an experiemnt
i am yet to implement this function
The idea is that the AWS env is meant for testing ideas or if there is an active connection to them means i might be running some kind of test or taking logs or something
The plan to do this is similar to how i built my WireGuard tracking system
it basically checks `wg show` and sees the networks transfer dataI
Database-Driven Configuration
Similar to Windows Registry for key-value management, I use a database to store:
Instance lists for automated shutdown
Sleep detection thresholds
Cost alert parameters
Environment-specific configurations
This enables runtime parameter adjustments without code changes.
Why n8n Crushes Lambda for R&D Automation 🚀
Lambda Reality Check: Per-invocation pricing + cold starts + debugging nightmares
n8n Advantages:
Fixed cost (its self-hosted, you can't beat that AWS) vs per-execution pricing (99% cheaper for frequent workflows)
Visual workflow debugging that actually helps troubleshoot
No cold starts for consistent performance
Webhook reliability without the serverless complexity
For daily cost notifications and automation workflows, n8n delivers enterprise-grade functionality at a fraction of Lambda's cost.
The Business Impact: Numbers That Matter 📊
Cost Optimisation Results
24/7 running instances
Always on
Sleep-based shutdown
60-80%
Cost visibility
Monthly surprises
Daily tracking
Immediate
Management overhead
Manual processes
Automated workflows
25+ hours/month
Workflow costs
Lambda functions
n8n automation
99% cheaper
Technical Advantages
Hybrid Intelligence: Home lab handles persistent workloads, AWS scales for experiments
Presence-Based Resource Management: Infrastructure mirrors your actual work patterns
Real-Time Cost Attribution: Every experiment's financial impact is immediately visible
Secure Network Integration: Private connectivity without compromising security
Implementation Insights: What Actually Matters 🎯
Cost Monitoring Setup
Enable Cost Explorer API with minimal permissions
Deploy n8n workflow for daily cost tracking
Configure Discord webhooks for instant notifications
Set up attribution logic to track experiment impacts
Infrastructure Automation
Deploy VPC with public/private subnet architecture
Configure the bastion host with presence-based auto-startup
Set up sleep detection automation with database configuration
Security Implementation
IP-restricted access from home DC locations
Audit-ready connections for compliance requirements(yeah, what better than saying I have audit logs for my homelab, maybe even get an ISO27001, cause why not)
Key Takeaways for Technical Leaders 💡
Cost Visibility = Control: The $0.01/day investment in cost monitoring prevents $1000+ monthly surprises
Automation Prevents Waste: Sleep-based shutdown saves 60-80% on compute costs without impacting productivity
Hybrid Approach Maximises Value: Home lab for persistence, AWS for scalability, VPN for security
n8n > Lambda for Workflow Automation: Fixed pricing, visual debugging, and no cold starts win for R&D workflows
Presence Detection Eliminates Manual Management: Your infrastructure should adapt to your work patterns, not the other way around
The Bottom Line 🏆
This architecture transforms R&D experimentation from a cost centre into a controlled, measurable, and highly automated technical asset.
The system knows when I'm home, when I'm sleeping, and exactly what each experiment costs. Infrastructure starts up as I walk in the door and shuts down when I'm inactive—all while maintaining enterprise-grade security and providing immediate cost feedback.
Result: Maximum experimentation velocity with zero waste and complete cost transparency.
Your R&D environment should serve your curiosity, not drain your budget. This setup delivers both.
Enterprise Application: Smart Cost Control for Development Teams 🏢
For businesses with development teams, this system automatically shuts down infrastructure when developers aren't actively using it. The beauty lies in the database-driven override mechanism—unless a developer explicitly changes the shutdown prevention value to 1 in the database, the system will power down non-critical services when the team isn't in the office.
This creates a default-to-savings approach while giving developers full control when they need 24/7 uptime for critical testing or deployment cycles. No more forgotten instances burning budget overnight, yet no friction when teams genuinely need persistent infrastructure.
Want to build something similar? Start with cost monitoring, add presence detection, and let automation handle the rest. The future of R&D infrastructure is intelligent, hybrid, and cost-aware.
Last updated