Agent Architecture
For senior engineers building AI systems
How production AI systems fail, scale, and recover.
Read deep engineering essays and guided paths, prepare for Staff and Principal interviews, and explore an upcoming live cohort on applied AI engineering.
By Amit Kumar, Principal Software Architect
What you can do on ArchCrux
Learn
Read engineering essays and follow guided paths through production AI systems.
Prepare for engineering interviews
Reason through production failures and system design decisions.
Distributed Systems & System Design →Databases, networking, consistency and reliability.
AI Engineering →Agents, retrieval, inference and evaluation.
Featured engineering resource
All resources →Field Guide 01 · Free PDF
Context Engineering: What should your AI actually see?
More context isn't always better. Learn how to select, structure, and evaluate the information your LLMs and agents receive.
View the field guide →10-page PDF · 201 KB
Interview Prep
Questions from the Interview Prep library
Retrieval and RAG
The query rewrite removed a version number. Why did RAG still look healthy?
Model and Inference Engineering
KV cache is nearly full and p99 doubled. What would you investigate?
Interview Arena · Coming Soon
A separate app for live AI interviews with adaptive follow-up questions.
Engineering reading
Latest essays
Agents
How Coding Agents Actually Work: Inside the Agent Harness
A technical look inside coding agent harnesses, covering context management, repository search, tools, sandboxes, subagents, durable sessions, recovery, concurrency, completion and evaluation.
Production AI Systems
OpenAI Decisions API vs Jev: Designing Systems Around Decisions Instead of Generation
A technical comparison of OpenAI Decisions API and Jev, and how to design a decision layer for routing, tool selection, risk, abstention, calibration and AI agent control.
Agents
Designing Containment for Production AI Agents
Production AI agent containment needs more than a sandbox. It needs separate information, execution and effect boundaries, scoped capabilities, egress controls, quarantine, and independent runtime enforcement.
Learn live · Program details
Applied AI Engineering Cohort
A live online course for software engineers designing LLM applications and agent systems from first principles.
6 live sessions3 weekends18 live hoursOnline
The curriculum is available to review. Dates and enrollment details are being finalized.
View Cohort →
Amit Kumar
Principal Software Architect
Writes about distributed systems, AI systems, and production judgment.