Writing

Notes on data engineering, cloud & AI

Long-form series and practical articles on building data platforms that transfer across tools — written from hands-on delivery, not vendor marketing.

Series · 14 parts · Newest

SQL for Every Data Platform

The first skill any new platform assumes: SQL. Fourteen parts, fundamentals to procedural SQL — schema design (star vs. snowflake), data cleaning, semi-structured/JSON data, the query cookbook, Slowly Changing Dimensions & incremental loading, performance tuning, simulating a platform locally with dbt, stored procedures/triggers, and a "same query, every platform" comparison across BigQuery, Snowflake, Databricks, Fabric, and Redshift. Code companion: sql-for-data-platforms on GitHub, 14 runnable notebooks.

July 2026 · 14 parts Start the series →
Series · 12 parts

Learn the Pattern, Not the Product

Platform-agnostic data engineering fundamentals that transfer across Snowflake, Databricks, BigQuery and Microsoft Fabric. Twelve core concepts, each explained "in 60 seconds" and then mapped — in a "same pattern, every platform" table — so the knowledge moves with you instead of expiring with a vendor's UI.

June 2026 · 12 parts Start the series →
Series · 4 parts

The Modern Data Ecosystem

A practitioner's guide to the three specialized engineering roles shaping data teams: Data Engineer, Data Architect, and ML/AI Engineer. Architecture diagrams, tool-stack breakdowns, and comparative tables for each role.

June 2026 · 4 parts Read the series →
Guide

Tackling the Data: Descriptive Analytics

A walkthrough of the key steps in descriptive analytics — from data collection and cleaning to transformation, visualization, and sharing results — with diagrams and an animated process flow.

July 2025 Read the guide →
Types of Analytics · 4 parts

Visualization Analytics

The cross-cutting "how" of every analytics type: choosing the right chart from the question, perceptual encoding (why position beats colour), advanced charts, and dashboard design. Four chapters, every chart rendered by the Python shown, on the shared retail dataset.

June 2026 · 4 parts Start the series →
Types of Analytics · 4 parts

Diagnostic Analytics

Why did it happen? From a symptom to tested causes — correlation-to-target, t-tests and chi-square, disciplined drill-down, and the discipline that correlation is not causation. Four chapters with real, rendered statistics (t = −55.6, χ² < 0.001).

June 2026 · 4 parts Start the series →
Types of Analytics · 4 parts

Predictive Analytics

What will happen? Framing (regression / classification / clustering), the train-test split and overfitting, and the metrics that keep you honest (R²/MAE, precision/recall, ROC). Four chapters with scikit-learn models trained and scored on a held-out set.

June 2026 · 4 parts Start the series →
Types of Analytics · 4 parts

Prescriptive Analytics

What should we do? What-if scenarios, business rules and constraints, and budget optimization under limits with SciPy linprog — plus the discipline of feasible-not-just-optimal. Four chapters built around one real allocation decision.

June 2026 · 4 parts Start the series →
Capstone

From Notebook to Platform Pipeline

The bridge from "works on my laptop" to a scheduled pipeline on Fabric / Snowflake / Databricks / BigQuery — without rewriting the logic. Modularize, parameterize, package, orchestrate, and a same-pattern-every-platform table. Ties the library to the Learn the Pattern series.

June 2026 Read the guide →