AK-DataGeek is a boutique data engineering and AI consultancy. We migrate legacy data warehouses to Snowflake, Redshift and BigQuery, design lakehouse architectures, and ship production AI systems with Claude Code and agentic workflows. 13+ years across aviation, insurance, transport and fintech.
Legacy Oracle, Teradata, SQL Server and PostgreSQL migrations to Snowflake, Amazon Redshift, or Google BigQuery. End-to-end ownership: discovery, schema redesign, ELT pipelines, cutover and post-migration validation. Specialised in regulated industries where downtime carries real penalties.
Greenfield warehouse and data lake architecture using Kimball, Data Vault, or hybrid bitemporal approaches. Slowly-changing dimensions, partitioning strategies and cluster keys engineered to survive 10× data growth without rewrites.
Production ML pipelines, signal computation engines, real-time inference layers. Bitemporal feature stores. Customer-facing AI products with backtest-ready architectures. From research notebook to GCP Cloud Run in weeks, not quarters.
Claude Code workflows, MCP server integration, custom agentic systems for data operations. Replace manual data wrangling with AI agents that ingest, validate, and route. Tooling that turns a single senior engineer into the output of a team.
Lakehouse architecture on Databricks — Delta Lake design, Unity Catalog governance, Spark optimisation, and migration from legacy Hadoop or cloud warehouse stacks. Delivered in partnership with a certified Databricks architect. From proof-of-concept to production-grade data platform.
Client names are kept confidential under NDA. Engagement patterns and outcomes below describe the shape of work we deliver — composite of multiple projects across regulated industries.
Modern data work isn't done by hand anymore. AK-DataGeek leans heavily on AI-augmented workflows — Claude Code for engineering, MCP servers for tool integration, OpenRouter for cost-optimised model routing — to deliver in weeks what traditional teams take quarters to ship.
What this means for clients: shorter cycles, lower cost, higher quality. A migration that would take a 3-person team six months gets done in eight weeks, with better test coverage and cleaner architectural documentation.
AI doesn't replace senior judgment — it amplifies it. No code ships unreviewed. Every commit goes through human review, every architectural decision lives in an ADR, every customer-facing output is verified against source. That's the line.
Every line of code, every architectural decision, every client-facing artefact comes from senior hands. We don't sell juniors at senior rates. The trade-off is scale — we take one or two engagements at a time and stay deep in them. Most of our clients prefer this trade.
Discovery, design, build, deploy, validate, document — all under one roof. We don't hand off between specialists. The same person who designs the Snowflake schema writes the dbt models, configures the Airflow DAG, and explains the trade-offs to your CTO. This eliminates the most expensive failure mode in consulting: lossy handoffs.
Every architectural decision lives in an ADR (Architecture Decision Record). Every schema change has a migration script. Every data contract has a test. When the engagement ends, your team owns a system they can run, debug and extend without us. That's the whole point.
We use Claude Code, MCP servers, and custom agents heavily — but every artefact is human-reviewed before it ships. AI accelerates the routine 70% of engineering work; senior judgment owns the 30% that matters. Clients get faster delivery without inheriting opaque AI-generated code they cannot understand or maintain.
Insurance, banking, transport, regulated tech — sectors where data correctness, audit trails and compliance are not optional. Our architectural defaults assume regulators will read the audit log. Bitemporal modelling, point-in-time correctness, full lineage and test coverage are baseline, not premium add-ons.
AK-DataGeek is a boutique data engineering and AI consultancy. We specialise in four practices: cloud data warehouse migrations from legacy systems like Oracle, Teradata and SQL Server to modern platforms such as Snowflake, Amazon Redshift and Google BigQuery; data warehouse and lakehouse design using Kimball and Data Vault methodologies; production AI and machine learning systems; and agentic AI workflows using tools like Claude Code and MCP servers.
Our team has delivered data engineering work for enterprises across regulated industries including aviation (British Airways), insurance (Aviva, Cuvva), transport (Transport for London, Arriva Group), fintech (Capitalise.com), logistics (UPS), engineering (BRE Group), and through technology consultancies (CodiLime). Specific project details are kept confidential under client NDAs — case studies on this site present the shape and outcome of work without naming individual engagements.
Core stack covers Python, PySpark, T-SQL, PL/SQL and ANSI SQL. Cloud warehouses: Snowflake, Amazon Redshift, Google BigQuery, plus legacy Oracle, Teradata and PostgreSQL. Cloud platforms: AWS and Google Cloud Platform. Orchestration: Apache Airflow, dbt, Cloud Composer. AI and ML: Claude Code, PyTorch, YOLOv8, Weights & Biases, DVC. Agentic layer: MCP servers, OpenRouter, Perplexity, Gemini, custom agent loops.
AK-DataGeek is a boutique consultancy that uses AI-augmented workflows to deliver senior-level work at a fraction of traditional consulting timelines. By leveraging Claude Code, MCP servers, and custom agent systems, we produce the output of a small team — with shorter delivery cycles, lower cost, and clean architectural documentation. The model trades scale for speed and depth. Clients work directly with senior engineers, never with junior staff billing at senior rates.
AK-DataGeek is based in Krakow, Poland and operates remote-first. Engagements are accepted across Poland, the United Kingdom, Germany, and the broader European Union. English-speaking engagements globally are accepted on a selective basis. Both Polish and English are working languages.
Yes. AK-DataGeek currently runs two production AI products. Nestis.cloud is a compliance-first SaaS infrastructure for regulated intermediaries, selected as a 2026 EEC Startup Challenge semi-finalist in the 4TECH category. Straitsmonitor.com is a commodity flow intelligence platform serving trading desks. Both products demonstrate our ability to ship production AI systems end-to-end.
Contact AK-DataGeek via email at hello@ak-datageek.com or through LinkedIn. Engagements are accepted selectively — typically one or two projects at a time — with availability from Q3 2026. The first conversation is free and exploratory; we'll figure out fit before discussing scope or pricing.
Available for selective consulting engagements from Q3 2026 — DWH migrations, AI/ML systems, agentic automation. One or two projects at a time, deeply hands-on. Drop a line if that fits how you work.
Send a message