Kubai Kevin

Backend engineering, AI tooling and developer careers

Explainers on backend engineering, AI tooling and developer careers: mechanisms, failure modes and decision checklists. Drafted with an LLM pipeline; figures are illustrative.

Articles are drafted by an automated LLM pipeline under the editorial direction of Kubai Kevin, a software developer based in Nairobi, Kenya. Topics are drawn from a curated list of recurring backend, AI and career themes; the pipeline is gated by hard quality controls (minimum length, duplicate detection, boilerplate rejection). Individual line-by-line human review is not performed on every post before publication — the author maintains the quality gates and corrects articles when readers report issues. Full AI content policy → · Editorial process →

Latest Posts

CPU-based autoscaling is mostly wrong

CPU-based autoscaling under-provisions when work is blocked, not computing.

14 min read

LLMs Kubernetes Serverless

Graceful shutdown: SIGKILL is not the enemy

Graceful shutdown for long-running workers fails when jobs outlast the grace period. Add a deadline check and checkpointing to stop losing work on deploy.

14 min read

Kubernetes Distributed Systems

CSRF and SameSite cookies: what breaks

CSRF and SameSite cookies: why Lax still allows state-changing GETs, how to add a double-submit token, and the Secure flag mistake that silently drops…

13 min read

Security Payments

On-call without derailing projects

On-call rotations erode project delivery when alerts are noisy and capacity isn't planned.

13 min read

Reliability

Object storage lifecycle policies explained

Object storage lifecycle policies: avoid early deletion fees, orphaned uploads, and silent data loss.

14 min read

AWS Azure Google Cloud

Senior without the title: what stalls engineers

Senior work without the senior title: how freelance engineers build real technical depth through decision logs, timezone-aware design, and payment…

16 min read

Payments Kubernetes Reliability

Security engineering isn't a ladder

Security engineering careers aren't a ladder—they're a lattice.

13 min read

LLMs Security Reliability

RBAC vs ABAC: pick by blast radius

RBAC vs ABAC for services: why role explosion and attribute spoofing fail in opposite directions, with runnable policy code and a deny-on-missing test.

12 min read

AI Agents Identity & Access

Kubernetes Requests and Limits: Seven Production Traps

Why pods get OOMKilled, throttled, or evicted: seven common requests-and-limits mistakes, how to detect each with real metrics, and which to fix first.

15 min read

Java Kubernetes Containers

Human-in-the-loop: when agents need a human veto

Design agent approval gates that live in the executor, not the prompt: signed single-use tokens, per-action scope, audit trails and anti-rubber-stamp checks.

14 min read

LLMs Payments AI Agents

Multi-region active-active: the hidden costs

Active-active across AWS, Azure and Google Cloud is a distributed-systems build, not a product. Here are its real costs and failure modes.

17 min read

AWS Azure Google Cloud

Python deps: uv vs Poetry vs pip-tools

Compare uv, Poetry and pip-tools by resolver, lockfile and environment model, with a worked example, failure modes and a decision checklist.

12 min read

AWS Python Containers

Your platform team is too big

Small engineering orgs often staff a platform team before scoping the platform as a product. Here is a staffing test, a written boundary, and paved roads.

18 min read

Terraform Reliability Engineering Careers

Freelance vs full-time: African devs' real tradeoffs

A structural comparison of freelance contracts and full-time remote jobs for African developers: risk transfer, utilization, tax, IP, and a worked pay comparison.

12 min read

Payments Africa Tech Mobile Money

Distributed locks: safe only with fencing

A lease can expire while its holder is still running. Fencing tokens, enforced by the protected resource, are what make a distributed lock safe.

17 min read

AWS Redis PostgreSQL

Virtual threads: what actually changes

What virtual threads change in a blocking Java service, where the real bottleneck moves, how pinning and thread-locals bite, and how to measure it.

15 min read

Java Redis Messaging

Structured logging: searchable at scale

Why free-text logs get slow and expensive at scale, and how a stable schema, shared logger, and value-based retention fix both problems.

20 min read

Python Google Cloud Observability

7 ADR traps that cause stale decisions

How architecture decision records go stale: template bloat, drifting status fields, hidden storage, broken numbering, missing context, no workflow trigger, no review.

15 min read

AI Agents PostgreSQL Software Architecture

Model evals: the gaps that let drift slip through

Aggregate eval scores hide subtle model drift. Learn per-example diffing, slice reporting, pinned baselines, and a minimal CI gate that catches behavior changes.

15 min read

LLMs MLOps AI Agents

Canary rollbacks: automate the decision, not the deploy

Design canary analysis that compares against a live baseline, guards on sample size, and avoids the false positives that erode trust in automation.

13 min read

CI/CD Observability

Open source contributions: the credibility trap

Why merged pull requests often fail to persuade hiring teams, and how to choose projects and shape contributions that read as engineering judgment.

18 min read

Engineering Careers

Saga pattern: orchestration vs choreography in payments

How payment sagas coordinate steps, why refunds are not charge reversals, and a decision framework for choosing orchestration or choreography.

16 min read

Payments Identity & Access Distributed Systems

WAF and bot management: what they miss

WAFs and bot managers filter known-bad requests, not business logic. Map blind spots, tune rules safely, and add application-layer controls.

16 min read

AWS Security AI Agents

FastAPI vs Django vs Flask: API picks for 2026

Compare FastAPI, Django, and Flask on concurrency, validation, ORM, auth, and operations, with a worked latency example and a 30-minute check.

15 min read

LLMs Python API Design

Browse older posts →