For senior engineers building AI systems
Deep engineering essays on RAG, agents, LLM infrastructure, and the distributed systems behind them. Focused on the architecture decisions that determine whether these systems work in production.
By Amit Kumar, Principal Software Architect
Scope
01
RAG, agents, evaluation, permissions, engineering knowledge, and generation control.
02
Retries, rate limits, multi-tenancy, tail latency, scheduling, cost, and operational failure.
03
Correctness boundaries, tradeoffs, failure recovery, and the decisions hidden behind architecture diagrams.
From the journal
Jev is useful when an AI system needs fast bounded decisions around generative work. The production design still needs state construction, versioned questions, policy, calibration, observability, and clear boundaries around when to act, escalate, or collect more evidence.
Read essayProduction Agent Skills are deployments, not files. They need immutable versions, controlled rollout, revocation, trust boundaries, progressive resolution, and a runtime path designed for scale.
Read essayRuntime authorization for AI agents needs more than an ALLOW decision. Production systems need exact action identity, bounded permits, reservations, revocation, serialization, and reconciliation with the resulting business effect.
Read essay