ANTHONY PENA  ·  DATA ENGINEER  ·  BUILD IN PUBLIC

ENGINEERING
AS
DOCUMENTATION

Architecting high-velocity data pipelines. No black boxes. Every schema, stream, and anomaly documented in real time across PostgreSQL, Redpanda, RisingWave, ChromaDB, all self-hosted, zero cloud dependencies.

ENTER THE CODEX RECENT ENTRIES →
3
SYSTEMS SHIPPED
92
TESTS GREEN
19
CONTAINERS
59
LOG ENTRIES
SYSTEM: FRG_BLOG_v2.0
STATUS: ● LIVE
SOURCE: MDX
// SYSTEM TELEMETRY
UPLINK ● STABLE
TEST SUITES 92/92 GREEN
CONTAINERS 19 ACTIVE
CB DETECTION <500MS
NULL_RATIO 0.4% OK
ENTRIES 59 LOGGED
// SIGNAL

// THE SYSTEMS

01

PROJECT CODEX · ANALYTICS WAREHOUSE

SPECTRUM ● OPERATIONAL 32/32 TESTS

A self-hosted Amplitude built entirely on PostgreSQL 16 and async Python, proving a well-designed schema can replace an analytics compute stack. Clickstream ingestion under 5ms, idempotent ELT into a partitioned star schema, sessions and funnels computed in pure SQL, auto-provisioned Grafana dashboards.

INGEST
──→
STAGING
──→
STAR SCHEMA
──→
ANALYTICS MVs
PostgreSQL 16 · FastAPI · Grafana
Codex 100% written · 7 build logs Last updated Jun 11, 2026
02

PROJECT CODEX · AI OBSERVABILITY

PHRONIS ● OPERATIONAL

Real-time observability and circuit breaking for AI agents. Streaming SQL detects a runaway agent in under 500ms and halts it, while batch tools are still waiting for their next aggregation window.

AGENTS
──→
REDPANDA
──→
RISINGWAVE
──→
BREAKER
11 containers · k8s
Codex 100% written · 7 build logsLast: Jun 05
03

PROJECT CODEX · SEMANTIC SEARCH

CONTEXTFLOW ● OPERATIONAL

ETL pipeline that turns 100 multilingual PDFs into a semantic index. Ask a question in plain English, get ranked answers with page-level provenance in under 2 seconds, fully local with zero API cost.

EXTRACT
──→
EMBED
──→
CHROMADB
──→
QUERY
60/60 tests · Airflow
Codex 100% written · 6 build logsLast: Jun 02

// NOW ON YOUTUBE

// LATEST LOG

Building a Brain to See What Wakes Up

AUG 05

READ THE ENTRY →

// REFERENCE INDEX — CONCEPTS

REF_04 · BUILD · KLAYEN
We Are Building Our Own AI Agent Platform
JUL 25 · 7 MIN›
REF_03 · POSTGRESQL
REFRESH MATERIALIZED VIEW CONCURRENTLY: How It Works, How to Break It
JUN 19 · 5 MIN›
REF_02 · STREAMING SQL
Incremental Materialized Views, Explained
JUN 08 · 5 MIN›
REF_01 · DIMENSIONAL MODELING
Star Schema vs Snowflake — When to Actually Use Each
JUN 07 · 6 MIN›

// RECENT LOG

GROWTH GROWTH · DIGITAL CONSCIOUSNESS

Building a Brain to See What Wakes Up

This week I gave my code and my life a shared memory. That's the tools post. The real question underneath it is whether a mind built this way could ever have an interior of its own, and I want to argue three positions on that, in the open, with everything I've read to back them.

AUG 05, 2026 · 32 MIN READ ›
CONCEPT BUILD · KLAYEN

We Are Building Our Own AI Agent Platform

On why Flowise and n8n weren't enough, and what we decided to build instead.

JUL 25, 2026 · 7 MIN READ ›
GROWTH GROWTH · BUILD IN PUBLIC

Frankl and the Creative Blockade

On reading Man's Search for Meaning and what it taught me about the gap between having everything ready and not being able to start.

JUL 14, 2026 · 5 MIN READ ›
SYSTEM SYS · TOOLING

We Migrated from Node.js to Bun and It Just Disappeared

What a monorepo migration actually looks like, what broke, what didn't, and why the best tooling is the tooling you stop thinking about.

JUL 14, 2026 · 6 MIN READ ›
SYSTEM BUILD · DESIGN SYSTEMS

Building Klayen: What Neumorphism Actually Requires

A premium agency site rebuild with a custom neumorphic design system. The shadow constraint, the scroll-driven animation pattern, and why I kept the inline styles.

JUL 03, 2026 · 9 MIN READ ›
GROWTH GROWTH · REFLECTION

Venezuela, June 2026

A tribute to those lost in the twin earthquakes that struck La Guaira and Caracas, the pain of watching your people suffer, and why helping is the only answer we have left.

JUN 26, 2026 · 6 MIN READ ›
SYSTEM CORPUS · BUILD

I Saw Microsoft's MarkItDown. Corpus Is What I Built Instead

A free web tool to convert PDF, DOCX, and HTML to clean Markdown. No Python, no install. Built as a deployable alternative to Microsoft's MarkItDown.

JUN 23, 2026 · 9 MIN READ ›
GROWTH GROWTH · BUILD IN PUBLIC

The first one

I didn't wait until I was ready. I decided, and then I moved.

JUN 20, 2026 · 3 MIN READ ›
FACT
DIM_A
DIM_B
CONCEPT CONCEPTS · POSTGRESQL

REFRESH MATERIALIZED VIEW CONCURRENTLY: How It Works, How to Break It

The incremental refresh, and the one mistake that turns it into a full rewrite.

JUN 19, 2026 · 5 MIN READ ›
VIEW ALL 59 ENTRIES →

// WHAT YOU'LL FIND HERE

SYSTEMS

Project
Codexes

Shipped, working systems, each with a full build log. Architecture decisions, anomalies caught in production, fixes deployed. Three systems live: an analytics warehouse, an AI observability stack, a semantic search pipeline.

PostgreSQL Streaming SQL Vector Search Podman

CONCEPTS

Engineering
Explained

Data engineering concepts and industry signal, explained through real implementations, not textbooks. Star schemas, streaming aggregation, idempotency, embeddings. From systems that actually run.

Star Schema Streaming Embeddings Idempotency

GROWTH

Process
Architecture

The human side of building alone. Deep work and the discipline it asks for, what failure actually teaches, why I write before I have the answers. This is where I think out loud about becoming the engineer and the person this work requires. Here, and on YouTube.

Process Discipline YouTube Build Logs
// DOCUMENTING THE WORK — AND THE HUMAN BEHIND IT — SINCE 2021
ANTHONY PENA

I build data systems and write about everything around them: the architecture, the failures, the long detours, and what each one teaches me.

I've been documenting in public since 2021, before I had anything technical to show, because the process is worth showing, not just the result. Stay a while and you'll find more than pipelines here. You'll find the person building them.

Engineering as documentation. Life as documentation. To me they were always the same instinct.

OPERATOR PROFILE →