SERVICES · QA ENGINEERING

Build the right
QA architecture.

Every team's QA problem is different. I start by learning yours, then building the architecture that fits. Recent focus has been on teams shipping AI-drafted code, where traditional QA practices no longer fit and the answer is a different methodology entirely.

see work get in touch

01 / what I help with

I work across most QA areas: strategy, test design, manual testing, automation, platforms, review practices. The three focus areas below reflect where recent work has concentrated, driven by how AI-enabled development is changing the way teams ship.

QA Architecture & Strategy

Test architecture for teams where coupled scripts have hit the wall. Recent work: HTTP-first invariant frameworks with test-account pool isolation, an agentic QA tester that derives scenarios from specs and executes across blackbox / DB / API layers, and self-healing preflight using consensus voting across parallel LLM calls. The common thread: test rules, not their implementation.

Test strategy & planning
Test architecture design & review
Invariant test frameworks (HTTP + API)
Agentic QA (spec-driven, disposable)
Self-healing preflight (consensus voting, auto-PRs)

QA Orchestration & Platform

Test frameworks get robust once there's proper orchestration underneath. Queueing, test-account locking, manifest management, Jenkins integration, failure triage, and dashboards that surface the operating model. Infrastructure that makes parallel runs reliable and failures actionable.

Orchestration backends (Node.js, Postgres, BullMQ)
Test account & resource locking
Manifest ingestion (versioned test catalog)
Jenkins / CI integration, deploy-to-test mapping
Failure triage with AI analysis
Operating-model dashboards (Next.js)

AI & Full-Stack Systems

When the QA work calls for building the systems underneath, I also do that. REST APIs, multi-tenant platforms, RAG pipelines with grounding verification, document intelligence.

REST / OpenAPI specs, compliance-first
Multi-tenant RBAC, OAuth2 / JWT
Background job processing (pg-boss, BullMQ)
RAG pipelines with grounding & HITL
Document extraction (OCR, vision)
Next.js 15 apps, strict TypeScript

02 / how I work

The method underneath the three services above.

Start with the problem. Every engagement begins with the team, the app, the cadence, the risks. No template, no preconceptions.
Design what fits. Choose the testing layers, the automation boundary, the methodology. Architecture shaped by your situation, not a reference implementation.
Test rules, not flows. Where it fits: invariants over UI scripts, HTTP-first, decouple tests from implementation. Where UI matters, self-healing or agents fill the gap.
Orchestrate at scale. Queues, locks, manifest, triage. Run a hundred tests in parallel and trust the results.

see selected work → read notes from the build → get in touch →