Articles
Four Properties of Generative AI: A Practical Framework for Calibrated Trust
A practical model for understanding next-token prediction, knowledge, working memory, and steerability—and choosing the right controls for reliable AI-assisted work.
AI Job Search — Telegram Notification Fork
A fork of MadsLorentzen/ai-job-search that adds an automated Telegram digest bot — scraping job portals daily, ranking matches, and pushing rich job cards directly to your phone.
Human Atlas: Offline Research Workspace Architecture
How Human Atlas ships a 12 MB standalone HTML atlas with embedded occupation packs, lazy-loaded society data, SHA-256 validation, and field-level evidence guardrails.
Human Atlas
Standalone research workspace for exploring life, work, and society contexts with visible evidence boundaries — built for product brainstorming and solo-founder research.
8 LeetCode Patterns to Master
A beginner-friendly map of eight reusable interview patterns — Two Pointer, BFS, Topological Sort, DFS, Top K, Modified Binary Search, Subset, and Sliding Window — with when-to-use rules, worked examples, and practice problems.
Go Remembers: Runnable Go Fundamentals
A self-registering Go learning repository where each concept file plugs into a sorted example runner via init() hooks.
Data Structures & Algorithms Learning Catalog
A Maven-organized Java repository of 70+ runnable DSA implementations indexed by category for progressive study.
Delivery Foundry: Governed Control Plane for AI Software Delivery
Architecture of a durable, resumable, evidence-verified control plane where a kernel owns state and side effects while a Plan Execution Coordinator proposes work under explicit policy envelopes.
Delivery Foundry: Autonomous Software Delivery Loop
Project overview of Delivery Foundry — a dual-track foundry for personal venture autonomy and organization 10x engineering, built on a shared governed kernel.
Portable AI Coding Knowledge with ARES
Project overview of ARES — the reference CLI for the AI Repository Standard that composes provider-specific artifacts from a canonical .ai/ source.
MD-AME
Autonomous YouTube production engine — dimension-parameterized media pipeline with PostgreSQL state machines.
Polymarket Agent
Autonomous prediction-market trading system — modular monolith with Postgres event bus, probability gates, and adaptive risk controls.
Building a Restartable Long-Video Processing Pipeline
How Shorts Factory uses a 16-stage deterministic pipeline with SQLite checkpointing to process hour-long videos into publishable shorts.
How to Prevent Contradictions in AI-Generated Documents
Section-sequential generation with cross-section coherence audits and guarded micro-fixes from A2A Brainstorm.
Designing a Deterministic Agentic Coding Orchestrator
Task-based agent loops with bounded retries, checkpoint rollback, and quality gates in skeleton-parallel.
Welcome to the AI Systems Engineering Knowledge Hub
What this knowledge base covers — systems architecture, production patterns, and lessons from real engineering work.
MD-AME: Autonomous Media Engine
Architecture of an autonomous YouTube production system built on deterministic pipelines and database-backed state machines.
Polymarket Trading Agent
Autonomous prediction market agent with signal processing, risk gates, and real-time capital protection.
Deterministic AI Pipelines
Why most AI pipelines fail in production — and how idempotent workers plus database state make them reliable.
Database-Backed State Machines
Using the database as the single source of truth to drive reliable workflow execution in distributed systems.
AI Orchestration Patterns
Central orchestrator patterns for coordinating AI workers, models, and pipelines without agent chaos.
LLM Guardrails
Engineering patterns for constraining LLM behavior — boundaries, validation, and rollback in production AI.