Technical Capabilities

The "I speak both languages" section

Not a Systems Administrator. Not a Software Engineer.

I'm an operations leader who bridges business requirements and technical execution.

What that means: I understand technical architecture and limitations. I can implement solutions directly when it's faster than delegating. I have technical conversations without translation layers. I can't build your entire infrastructure—but I can design operations around it.

Professional Tools

  • Salesforce
  • Zendesk
  • HubSpot
  • Looker
  • Power BI
  • SAP
  • Excel
  • Slack
  • Teams
  • TorchLMS

Personal Infrastructure (35+ Services)

  • Docker
  • Grafana
  • Prometheus
  • Loki
  • SQL
  • Vaultwarden
  • Forgejo
  • Borg Backup
  • Tailscale
  • Linux

Real Examples

Database Health Check

SQL
-- Vaultwarden password manager health check
-- Verify database accessibility and basic integrity

SELECT COUNT(*) as total_users FROM users;

SELECT
    COUNT(*) as total_ciphers,
    COUNT(DISTINCT user_uuid) as users_with_items
FROM ciphers;

SELECT
    datetime(updated_at, 'unixepoch') as last_activity
FROM users
ORDER BY updated_at DESC
LIMIT 1;
Simple queries to verify database accessibility for a self-hosted password manager. Not complex joins—just operational health verification that tells me "yes, the database is working."

Alert Threshold Tuning

PromQL
# Container CPU usage alert
# Trigger when any container exceeds 80% CPU for 5 minutes
rate(container_cpu_usage_seconds_total{name!=""}[5m]) > 0.8

# Memory pressure warning
# Alert when container memory exceeds 90% of limit
container_memory_usage_bytes / container_spec_memory_limit_bytes > 0.9

# Backup job failure detection
# Alert if backup hasn't succeeded in 25 hours
time() - backup_last_success_timestamp > 90000
Prometheus query language for monitoring infrastructure. These alert rules help identify problems before they become outages.

Error Pattern Detection

LogQL
# Find errors in Prometheus logs
{container_name="prometheus"} |= "error"

# Count error rate over time
sum(rate({container_name=~".+"} |= "error" [5m])) by (container_name)

# Filter for specific error patterns
{job="containers"}
    |= "error"
    != "context canceled"
    != "connection reset"
    | line_format "{{.container_name}}: {{.message}}"
Loki query language for log aggregation. Filter noise, surface real errors, calculate error rates from unstructured log data.

The Difference

At Work

Consumed data through interfaces built by technical teams. Used Looker dashboards, Salesforce reports, Power BI visualizations. Understood enough to ask good questions and request the right modifications.

At Home

Had to learn how query languages work because there's no IT department. Built the dashboards myself, wrote the alerts, designed the monitoring infrastructure. Trial by necessity.

Both Environments

Same analytical thinking—patterns, hypotheses, data validation. The syntax is different, but the mental model is identical: define the question, find the data, validate the answer.

The Value

Can have technical conversations without translation layers. Can implement solutions directly when appropriate. Can bridge business requirements and technical execution without being a bottleneck.

Applied Example: Alert Noise Reduction

Problem: Backup alerts constantly firing in my homelab. I learned to ignore them—not good. Real failures got missed in the noise. This experience led to building proper alert monitoring with RSS feeds and eventually a Discord bot for interactive monitoring.

1

Analyze Alert History

Pulled 30 days of alert data from Prometheus. Identified which alerts fired most frequently and which were actually actionable.

2

Identify False Positives

Discovered thresholds were set arbitrarily, not based on actual usage patterns. "Disk space > 80%" triggered constantly because 85% was normal operating range.

3

Adjust Based on Reality

Tuned thresholds based on actual baseline behavior. Added time-based conditions to prevent transient spikes from triggering alerts.

90% reduction in alert noise

Same approach used for: Conversion optimization, process improvement, operational troubleshooting—just applied to different problem domains.