Notes

Browse by topic

Three topic hubs; each one defines the area, offers a start path, and lists the notes.

Start here (reading paths)

Three short reading paths, one per topic.

Performance

1. TTFB and origin latencyUnderstand request phases and what “first byte” really contains.
2. Queueing basics and latency budgetsWhy tail latency explodes and how to set budgets that survive load.
3. Performance regressions checklistA workflow to confirm, isolate, and mitigate regressions.

Formal methods

1. Specs, invariants, and contractsWrite constraints you can review, test, and monitor.
2. Model checking primerExplore behaviour spaces and interpret counterexamples.
3. Session types and protocolsMake interaction structure explicit and checkable.

Distributed systems

1. Time, clocks, and orderingDefine “before” without relying on a single clock.
2. Failure modes, retries, idempotencyDesign operations so retry behaviour is safe and predictable.
3. Observability for distributed systemsDebug with evidence across service boundaries.

Latest notes

Six recent notes across all topics.

• Observability for Microservices: Unified Metrics, Traces and Logs in the Age of AIHow to build a unified observability stack for microservices that integrates metrics, distributed traces, and structured logs, including AI-assisted analysis patterns. performance
• Reducing Latency in Serverless and Microservice Environments: New Caching and Queueing StrategiesPractical strategies for reducing tail latency in serverless and microservice architectures, covering cold start mitigation, caching patterns, and queue-based load levelling. performance
• End-to-End Observability: A Must-Have for Multi-Cloud and Hybrid ArchitecturesHow to build observability that spans cloud providers and on-premises infrastructure, covering trace propagation, metric aggregation, and the organisational challenges of unified visibility. performance
• Scaling Formal Methods: Techniques for Integrating Verification into Industrial WorkflowsPractical approaches to integrating formal methods into real engineering workflows: lightweight specifications, incremental verification, and tool integration patterns. formal
• Reliable AI Requires Deterministic Foundations: Ensuring Data Quality and Context for LLMsWhy reliable AI system output depends on deterministic data pipelines, context management, and quality assurance, not just model capability. formal
• Formal Methods for Multi-Agent Protocols: Specifying and Verifying Communication PatternsHow to apply session types, process algebras, and model checking to specify and verify the communication protocols that multi-agent AI systems depend on. formal
• Observability for distributed systemsHow traces and correlation IDs turn incidents into evidence. distributed
• Performance regressions checklistA repeatable workflow for confirming and localising regressions. performance
• Specs, invariants, and contractsWhat to write down so behaviour stays reviewable. formal
• Failure modes, retries, and idempotencyWhy retries create duplicates and how to make outcomes safe. distributed
• Cache hierarchy: edge to originReason about caches as a system, not a single knob. performance
• Behavioural equivalence in plain EnglishWhat “same behaviour” means depends on your observations. formal

How notes are written

• Quick summaryA short statement of what the note covers and why it matters.
• Key ideasDefinitions and mini examples to build working intuition.
• Common pitfallsThe mistakes that recur in practice (measurement, assumptions, and semantics).
• Related notesInternal links that keep the knowledge base navigable and crawlable.