Sign In
Front Matter · #01

Introduction

Reading Progress4%

Introduction

By Joe Alongi

Welcome to my working notebook and companion repository. As 2026 began, I found myself immersed in a project that felt like the culmination of a decade's work. Instead of simply building another app, I was architecting a production-grade, full-stack telemetry and machine learning platform. This book acts as "The Book", documenting the entire journey. Each chapter provides comprehensive narrative deep dives (minimum 600 words), alongside code snippets and technology links.

Over the last ten years, my path has evolved from early web development, through the founding of startups, to deep software engineering and architecture. Those foundational years unlocked the paradigms I am now pouring into this platform. My goal here is simple: to champion the thoughtful coders. We're going to build this system together, starting from a completely fresh Mac install and working my way up to a deployed, secure, and observable ML-driven application. I will merge rigorous data engineering with the predictive power of machine learning, prioritizing quality and precision every step of the way.

For a brief summary of the platform's hypothesis, value add, architecture diagrams, and algorithms, please read the Whitepaper. For how the platform is operated in production (vendors, workflows, maintenance, contingencies), read Concept of Operations (CONOPS) or the operator quick reference docs/conops.md. For a visual slide-deck overview, see the companion Gamma presentation.

Note on Recent Evolution (2026 updates): The architecture now emphasizes Event Projections with production reliability features. Client commands route through Firebase Cloud Functions (ingestEvent, versioned), with Redpanda for events and Firestore (named "deml" database + dedicated security rules) for materialized read models. Django uses a Transactional Outbox (OutboxEvent + outbox_relay command) for reliable publishing. The telemetry_worker performs idempotent projections (with DLQ support). Firebase Functions and rules deploy via dedicated GitHub workflow. See updated diagrams; the loop is health-checked automatically by a synthetic probe in the telemetry worker and surfaced as the "Event Projections" component on the platform-status page.

Quick Links