Your data already knows the answers.
I help it speak clearly.
From raw, scattered metrics to precise, plain-English intelligence that drives real business decisions — without the usual noise.
Most teams drown in metrics. I build pipelines that collect cleanly, store efficiently (VictoriaMetrics), and let an LLM (DeepThought) explain what the data actually means — in plain English.
Example: “Why did our latency spike at 3am?” → DeepThought tells you the exact correlation with a database vacuum job, user load, and a failing cache instance. No more guessing.
Grafana panels designed for engineers and decision-makers, not slide decks. Purpose-built visualisations that surface the questions your team is really asking.
Statistical anomalies, subtle correlations, early warning signals — the stuff that hides in plain sight. I combine time-series analysis with LLM reasoning to surface insights humans would miss.
Your data becomes a living system: real-time alerts with plain-language explanations, predictive trends, and one-click drill-downs that actually make sense.
Custom collectors for MySQL, PostgreSQL, MSSQL, MongoDB, Elasticsearch, Kafka, AWS Athena, CSV — anything. Hot-reload config, concurrent workers, retry logic. I make sure nothing important is ever missed.
High-cardinality handling, smart schema design, long-term retention strategies, and efficient querying at scale. I make your metrics fast and cheap to store — forever.
Dashboards built for the people who read them. Custom panels, intelligent alerting, multi-tenancy, and visualisations that actually help humans make better decisions.
A local LLM layer that reads your time-series data and speaks plain English.
Before DeepThought
You see a spike in error rates at 02:14.
After DeepThought
“Error spike at 02:14 correlates with a 400% increase in slow database queries after a nightly ETL job. Root cause is likely lock contention in table X. Recommendation: add connection pooling or schedule the job 30 minutes later.”
35+ power metrics, temperature, humidity, UPS, chicken coop, Starlink — all feeding into VictoriaMetrics.
Correlated with brownouts and Starlink outages
Plain-English root cause analysis every morning
“Battery likely to fail in next 14 days”
I help teams move from “we have lots of metrics” to “we actually understand what’s happening and why”.
Remote-first, async-friendly, and obsessed with clarity.