The data pipeline lesson I learned after trying to force an ETL pattern into marketplace analytics.

The warehouse architecture lesson I learned after moving from ClickHouse to BigQuery, then back to ClickHouse again.

The product-market fit lesson I learned after Omnichannel moved from marketplace analytics into finance reconciliation.

My note on OLTP, OLAP, and why analytics workload needs a different kind of database.

Inspired by Austin Kleon's 'Show Your Work!', I'm starting to document my journey building an AI startup, sharing both successes and failures as I learn and grow in the AI and data space.

The meet with PydanticAI, which will be the fundamental of Mark AI system

Refactor backend from based on file type into modular

Initial deployment of Mark

The first day on building mark backend, build sandbox environment.

Decision to separate backend service from nextjs into fastapi

Consideration on how to build Mark as an agentic product

Mark focus on helping business user to by getting the context of their business connecting to their spesific data source

Mark, an AI Business Analyst that can become your 24/7 assistant