Mykhail M.

I build the clean, reliable data layer your team — and your AI tools — actually trust. From marketing, finance & CRM warehouses in BigQuery to real-time pipelines running at 200k+ events/sec, I've spent 15 years making data fast, accurate, and cost-effective. Most "data problems" aren't really about tools — they're about numbers finance can't trace, dashboards that silently break, and cloud bills that climb faster than the data does. I fix the layer underneath so the whole thing just runs — and often cut the bill while doing it. ⎯⎯⎯ What I do for you ⎯⎯⎯ ◆ Modern Warehouses in BigQuery (dbt + Medallion) — Unified, documented, AI-ready data from ad platforms (Meta, Google, TikTok), CRMs, and ERPs. For enhanza I consolidated financial data from 1,000+ organizations into one BigQuery platform with self-serve dashboards. ◆ AI-Ready & LLM Pipelines — Data modeled so your AI tools consume it directly: LLM enrichment/categorization, clean context windows, no manual prep. Shipped code-first with Claude Code + gcloud — no hand-written SQL in the console. ◆ Cut Your Cloud Bill (FinOps) — Runaway AWS costs are usually egress fees and over-priced managed services, not the data itself. I migrated a client's Dagster pipelines from AWS to Civo Kubernetes + Cloudflare R2 — eliminating S3 egress fees and slashing managed-cluster costs, with no orchestration rewrite. If your bill is climbing, I'll audit it and re-architect the expensive parts. ◆ Real-Time & High-Scale (when you need it) — Streaming in Flink/Kafka and sub-second OLAP in ClickHouse/StarRocks. I've run 200k+ ingest events/sec in an air-gapped cluster — so a standard pipeline is routine for me. ◆ Reliability & Observability — Monitoring, lineage, and schema-drift alerts so dashboards stop breaking. On one platform this cut data downtime by ~90%. ⎯⎯⎯ How I work ⎯⎯⎯ ▹ Root-cause, not patches. I fix the layer underneath so the same dashboard doesn't break again next month. Quick patches are how data platforms rot. ▹ Built to run without me. No "works once on my laptop" scripts — proper deployment, secrets, logging, alerts, and infrastructure as code (Terraform). Reproducible, and yours to keep. ▹ Straight about what I know. If something's outside my wheelhouse, I say so up front and give you the honest trade-off. I won't learn on your budget and call it experience. ▹ I own the outcome. Give me the goal and I'll work through the ambiguity and keep you posted — no micromanagement needed. ▹ A full team behind me. I run with a ~20-person team across data engineering, BI, DevOps, and AI/ML — so you're never betting on one person's calendar or blind spots. You work with me directly, and I pull in the right specialist when the scope calls for it. Senior ownership, agency depth. ⎯⎯⎯ Tech I work in daily ⎯⎯⎯ ▹ Warehouse & Analytics: BigQuery, dbt, dlt, Meltano, Medallion Architecture, Power BI, Looker Studio, Metabase ▹ Cloud & Orchestration: GCP, AWS, Azure, Civo, Airflow, Prefect, Dagster, Kubernetes, Terraform ▹ Streaming & Performance: Kafka, Flink, Spark/PySpark, ClickHouse, StarRocks, Iceberg, Delta, Polars, Cloudflare R2 ▹ AI-first delivery: Claude Code + gcloud/CLI, LLM enrichment pipelines ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ I don't just "install tools" — I design for compute cost (FinOps) and data reliability. If your cloud bill is climbing or your reporting can't be trusted, I'll audit the stack and tell you exactly what to fix. Want a quick, no-pressure look at your setup? Send me a message and I'll share where I'd start.