Alex Merced — Developer and technical writer sharing in-depth insights on data engineering, Apache Iceberg, data lakehouse architectures, Python tooling, and modern analytics platforms, with a strong focus on practical, hands-on learning.
610 articles from this blog
Table of Contents How to Evaluate a Family Instead of a Model OpenAI Anthropic Google Meta xAI The Chinese Open-Weight Labs Europe, and the Enterprise
Table of Contents Where the Specification Actually Stands Snowflake: Managed Storage, Open Catalog, Both at Once Databricks: Interoperability as a Com
Table of Contents Why the Gap Exists in the First Place What the New Architecture Actually Asks For Databricks: Put the Transactional Store Inside the
Table of Contents Four Assumptions Agents Break Where the Money Actually Goes Control One: Give Every Agent an Identity Control Two: Budgets and Limit
Table of Contents “Real-Time” Is Not a Specification Build the Budget Before the Pipeline Make Freshness Queryable Measuring Rather Than Estimating Th
Table of Contents What Actually Changed The Single-Node Engine Field Standing Up the Local Stack Writing and Reading, With Governance Intact The Spark
Table of Contents Three Properties People Treat as One What the Rules Actually Say in 2026 Minimum Sufficient Sovereignty Open Formats Are the Soverei
Table of Contents The Layer Cake and Its Interfaces Case One: Variant, and Why It Had to Live in Parquet Case Two: Geospatial, and the Statistics Prob
Table of Contents Why Language Models Fail at Tables What Gradient Boosting Got Right, and What It Costs The Mechanism: Prediction as In-Context Learn
Table of Contents What Maintenance Actually Is Why the Debt Accumulates Faster Than It Used To The Arithmetic of Write Amplification Why It Commoditiz
Table of Contents Six Architectures Wearing One Word Where the Bytes Actually Go Four Costs Nobody Puts in the Business Case Pushdown Is the Whole Gam
Table of Contents Filters Are Infrastructure, and They Have Two Jobs Radio: A Stranger’s Voice, and the First Modern Curators Television: Seeing Is Be
Table of Contents First, Why “Open” Is the Whole Point How the Stack Came Together Layer One: Apache Parquet, the File Format Layer Two: Apache Iceber
Table of Contents The Setup: A Shortage With No Villain and No Disaster A Short History of a Falling Price, and Why It Mattered Lens One: Supply, Dema
Table of Contents What Actually Happened, and Why Each Part Matters The Theory Under Indictment: Safety Through Restriction An Aside on the Broader Pa
Explores active analytics loops for AI agents that proactively monitor, investigate, and act on data anomalies beyond passive chat.
Explores Apache Polaris, an open source Iceberg REST catalog enabling multi-engine interoperability for open lakehouse architectures.
Explores the gap between AI agent demos and production-ready enterprise systems, focusing on data and knowledge challenges.
Explores Dremio's automated materialized views and reflections for optimizing AI-era lakehouse workloads beyond static cron-based maintenance.
Analysis of Dremio's Lakehouse AI report on shifting enterprise priorities from cost migration to agent-ready data platforms.
