Applied Systems
Engineering Blog
Technical post-mortems, benchmark telemetry, and incident runbooks authored by our principal consulting engineers.
Trending Blogs
- AI7 min
- Cloud8 min
Anatomy of a Production Kubernetes Cascading Failure: Post-Mortem & Fix
- AI10 min
Architecting Autonomous AI Agent Swarms with FastMCP & Structured Tool Routing
- Data12 min
Sub-Second Lakehouse Queries over Petabytes with Apache Iceberg and ClickHouse
All Engineering Runbooks (9)
Technical post-mortems, distributed systems runbooks, and AI architectures.
How Anthropic Built Claude's Multi-Agent Research System
Anthropic's Claude Research uses multiple AI agents that collaborate, reason, and coordinate in parallel, outperforming single agents by 90.2% on complex research tasks.
Anatomy of a Production Kubernetes Cascading Failure: Post-Mortem & Fix
A step-by-step breakdown of how a single misconfigured liveness probe triggered an avalanche of node evictions and a 38-minute outage across three availability zones.
Architecting Autonomous AI Agent Swarms with FastMCP & Structured Tool Routing
Moving beyond naive chat completions: how to orchestrate multi-agent state machines, deterministic tool schemas, and zero-latency context caching in enterprise workflows.
Sub-Second Lakehouse Queries over Petabytes with Apache Iceberg and ClickHouse
How to decouple object storage economics from real-time analytics velocity using open table metadata and vectorized columnar query engines.
Migrating to eBPF Zero-Trust Networking with Cilium and SPIFFE
Why traditional iptables-based network policies crumble at enterprise scale, and how eBPF enables kernel-level identity verification and mTLS without sidecar overhead.
How We Cut Enterprise CI Build Times from 48 Minutes to 3 Minutes
A deep dive into distributed remote caching, multi-platform container compilation, and Kubernetes-native ephemeral runner pools.
Architecting Hackathons as Real-Time Chaos Engineering Laboratories
Why Cerebral Hacks builds automated load generators, simulated infrastructure outages, and live scoring engines for our technical competitions.
PostgreSQL at 50,000 QPS: Indexing Tricks, Connection Pooling & Vacuum Tuning
Practical techniques for keeping relational databases responsive under extreme concurrent write amplification.
Threat Modelling Workshop Recap: Top 5 Vulnerabilities Uncovered by Attendees
A summary of the live architecture teardown from our July webinar, featuring real flaws discovered in distributed auth implementations.