Focus:
Time-series & observability
Stack:
VictoriaMetrics, Grafana, custom collectors
Availability:
Remote-first, async-friendly
Analytics & Observability
Most teams drown in metrics without getting closer to understanding them. I build pipelines that collect cleanly, store efficiently, and turn raw time-series data into plain-English answers rather than another dashboard nobody opens.
- Observability
- Time-series
- Statistical analysis
What this covers
Collection & federation
Custom collectors for MySQL, PostgreSQL, MSSQL, MongoDB, Elasticsearch, Kafka, AWS Athena, CSV and more, with hot-reload config, concurrent workers and retry logic.
VictoriaMetrics depth
High-cardinality handling, careful schema design, long-term retention strategy and efficient querying at scale.
Grafana that works
Dashboards built for the people who read them: custom panels, considered alerting, multi-tenancy.
Statistical & LLM-assisted analysis
Time-series analysis paired with LLM reasoning to surface correlations and early warnings a person would otherwise miss, the same underlying skill set behind DAS and Basilisk.
A real example
A rural home-monitoring setup feeding more than 35 power metrics, plus temperature, humidity, UPS state, a chicken coop and Starlink status into VictoriaMetrics: voltage drops correlated against brownouts and Starlink outages, plain-English root-cause summaries each morning, and predictive alerts like "battery likely to fail within 14 days" rather than a bare threshold breach.