Hi, I'm

Christoph Lengowski

QA for AI Systems | Product & Requirements

I combine QA leadership, AI quality, requirements engineering, and production-oriented delivery. I build systems, workflows, and assets that stay credible from the first requirement to release.

LLM QARequirements EngineeringProduct Delivery5+ Years Experience

Get in Touch View Projects Download CV

What I Bring

Experience at the intersection of quality engineering, delivery ownership, and AI-powered product development.

Quality Engineering

Building and managing QA processes in complex projects. Test strategy, test concepts, and quality reporting at management level.

Test Automation

Designing and implementing test automation frameworks with Playwright. CI/CD integration and sustainable test stability.

Product Development

Designed and built my own SaaS products. Architecture, UX, go-to-market - from problem to production-ready product.

AI & Automation

Integrating AI workflows into real products and processes. Prompt engineering, LLM-powered features, and intelligent automation.

Agile Delivery

Managing agile initiatives, release management, and stakeholder communication. Creating structure in dynamic environments.

Profile

About Me

I work at the intersection of QA leadership, AI quality, requirements engineering and hands-on product execution.

I am a Senior IT Consultant and QA Team Lead at a leading IT consulting firm. In complex project environments, I take responsibility for test strategy, test automation, and quality steering with a clear focus on delivery, risk management, and stakeholder alignment.

At the same time, I do not operate only at concept level: I design and build my own digital products, develop AI-assisted workflows, and turn vague requirements into implementable architectures, features, and quality mechanisms.

That gives me a combination many teams rarely get in one profile: QA leadership in demanding delivery contexts and hands-on product engineering with modern web, AI, and automation stacks.

Current focus

AI quality with measurable release gatesRequirements workflows for fuzzy product ideasProduction-grade web apps with Next.js and ReactReusable QA and delivery assets

Experience

5+ Years Delivery & QA

Role

QA Lead & Team Lead

Foundation

M.A. + Scrum/ISTQB

Builder profile

AI & SaaS Products

Quick profile

QA team leadership and delivery ownership in complex public-sector programs
Hands-on product engineering for AI-assisted SaaS and workflow products
Strong bridge between requirements, quality steering and technical execution
Focus on systems that stay reliable under real operating conditions

What teams value

I bring structure into ambiguous problem spaces and surface risks early.
I treat quality as part of delivery, not as a late control step.
I translate cleanly between domain experts, stakeholders and implementation teams.

Experience

Career milestones that shaped my profile.

2022 - Present

QA Team Lead

IT-Beratungsunternehmen (auf Anfrage)

Key Impact

QA leadership in a large public-sector program with team ownership, test strategy, and automation at delivery scale.

Leading a QA team with responsibility for test strategy, test automation, and quality management in a large-scale public sector project.

Built and led a QA team
Defined test strategy and test concepts
Implemented Playwright-based test automation
Set up Jenkins CI/CD pipelines and reporting
Stakeholder management at project leadership level

2020 - 2022

IT Consultant

IT-Beratungsunternehmen (auf Anfrage)

Key Impact

Built the foundation for delivery, requirements, and QA steering in real-world digitalization projects.

Consulting and hands-on work in digitalization projects. Test management, requirements engineering, and agile project management.

Test management with Jira, Xray, and Confluence
Requirements analysis and functional specification
Agile project management using Scrum
Quality assurance and acceptance testing

Projects

Selected projects across public sector delivery, QA leadership, and product engineering that show how I stabilize complex execution environments.

01Top Project

E-Gov Workflow Platform – Large-Scale QA in the Public Sector

QA Lead / Test Manager for a complex e-government platform

QA lead experience in a large-scale e-government program

Public sector delivery, team leadership, test strategy, automation, and release quality in one complex program.

QA team members led functionally

System domains covered

CI/CD

Automation anchored in Jenkins pipelines

Public Sector

E-government delivery in a highly regulated environment

A large-scale e-government platform for digital files and process handling in public administration. Within this complex program, I was responsible for planning, steering, and evolving the entire quality assurance setup, raising QA maturity both operationally and methodologically.

My role

QA Lead / Test Manager / IT Consultant

Tech Stack

Playwright · Jenkins · Jira · Xray

Challenge

The project ran in a highly complex public-sector environment with multiple clients, backend services, strong traceability requirements, and demanding release expectations. Quality had to be controlled not just through execution, but through team leadership, test strategy, defect management, and stakeholder reporting.

Solution

I built a structured test organization, led the QA team, and tightly connected test management, automation, and KPI-based reporting with engineering, project leadership, and client stakeholders. That turned quality into a controllable delivery capability instead of a reactive bottleneck shortly before releases.

Project context

Further development of a complex e-government platform for digital files and administrative case workflows
End-to-end QA responsibility across Web Client, Outlook Client, Admin Client, and backend services
Delivery in a public-sector environment with high expectations around quality, security, and auditability

Project scope

Functional leadership of a QA team of up to 8 people
Setup and steering of the full test management process
Creation of test concepts and test strategies
Planning, prioritization, and execution of release and regression testing
Ownership of the defect management process
Support for government-side test activities plus workshops and training sessions

Key achievements

Built a structured test organization inside a large e-government program
Introduced and expanded Playwright-based test automation
Integrated automated tests into Jenkins CI/CD pipelines
Improved release stability through systematic quality steering
Established test KPIs and reporting for leadership and stakeholders
Reduced friction in defect management through clear processes and prioritization

Tech Stack

PlaywrightJenkinsJiraXrayConfluence.NET / C#SQL ServerWebservicesSharePointOutlook Add-in

02Featured ProjectVisit Website

Nutrikompass

AI-powered SaaS platform for therapeutic nutrition planning

Tech Stack

Next.js · TypeScript · tRPC

Nutrikompass supports therapeutic residential groups in structured nutrition planning for residents with eating disorders. The platform combines clinical expertise with AI automation: a RAG-based knowledge base, an automated LLM evaluation framework with 5 clinical benchmarks, and two-stage prompt injection detection for safe use in care settings.

Challenge

Therapeutic residential groups face a recurring problem: nutrition planning for residents with eating disorders is time-consuming, error-prone, and barely standardized. AI support carries specific risks — from hallucinations to prompt injection — that are intolerable in clinical contexts.

Solution

A specialized SaaS platform with a multi-layer security architecture: RAG with pgvector for fact-grounded responses, LLM-as-a-Judge for automated quality scoring against clinical benchmarks, two-stage injection detection, and a full Playwright E2E test suite in the CI/CD pipeline.

Multi-tenant SaaS architecture with Next.js and Prisma
RAG knowledge base with pgvector and OpenAI embeddings
AI Evaluation Framework: 5 clinical benchmarks, LLM-as-a-Judge
Two-stage prompt injection detection for clinical safety
Playwright E2E tests + GitHub Actions CI/CD pipeline
Stripe subscription billing, NextAuth role management

Learnings

“Security in AI systems isn't a feature — it's architecture. In clinical contexts, every AI output must be traceable and verifiable. This fundamentally shaped my understanding of responsible AI integration.”

Tech Stack

Next.jsTypeScripttRPCPrismaSupabasepgvectorNextAuthStripeOpenAI APIFHIR ExportPlaywrightNetlify

Visit Website

03Featured Project

AI-QA-Framework

Requirements-driven QA framework for LLM systems with release gates

Tech Stack

Python · Pytest · Playwright

A Python-based open-source framework for production-oriented quality assurance of LLM applications. The focus is not just isolated tests, but a reliable QA workflow with 46 automated checks, traceability to quality requirements, RAG evaluation, multi-model support, and CI-backed release gates for AI features.

Challenge

LLM systems rarely fail because of a single bug. They fail through harder-to-control quality risks: non-deterministic responses, poor traceability, hallucinations, bias, prompt injection, and regression-prone UI flows. Classical assertions are not enough for that.

Solution

I built a requirements-driven testing framework across 7 quality dimensions: security, consistency, hallucination, performance, bias, RAG, and UI. It combines semantic evaluation, multi-provider tests for Claude, GPT, and Gemini, generic Playwright checks for chatbot interfaces, HTML reporting, and GitHub-Action-based quality gates before release.

46 automated tests across 7 quality dimensions for LLM risk areas
Requirements-driven QA with traceability and explicit release gates
Multi-model support for Claude, OpenAI GPT, and Google Gemini
8 RAG tests covering grounding, faithfulness, and contradictions
17 generic Playwright UI tests for chatbot interfaces
HTML reports and dashboard views for trends and regressions

Learnings

“AI quality becomes manageable only when testing, requirements, and release decisions are connected. Model quality alone is not enough; what matters is the discipline to make risks measurable and releasable.”

Tech Stack

PythonPytestPlaywrightClaude APIOpenAI APIGemini APIGitHub ActionsChart.js

GitHub

04Case Study

Guided Requirements Tool

AI-assisted requirements discovery with a structured interview flow

Tech Stack

Next.js 16 · React 19 · TypeScript

A full-stack product for product owners and business teams that turns vague ideas into usable user stories, acceptance criteria, NFRs, and an initial delivery plan within minutes. Technically, the tool combines a multi-step interview flow, Claude-generated summaries, parser-based result shaping, and PDF export inside a production-oriented Next.js architecture.

Challenge

The friction between business context and implementation is rarely about lack of expertise. Stakeholders know the problem, but often express requirements too vaguely or too inconsistently, which costs teams time, scope clarity, and quality.

Solution

I built a guided three-phase interview flow with context-aware follow-up questions and AI-assisted consolidation. Instead of dumping free text, the system produces structured artifacts: prioritized user stories, acceptance criteria, NFRs, open questions, and an initial sprint cut.

Next.js 16 app with React 19, TypeScript, and Tailwind CSS v4
Guided three-phase interview flow instead of blank requirement forms
Claude-generated phase summaries for built-in quality control
Generation of user stories, acceptance criteria, NFRs, and sprint order
Parser and result logic for reproducible structured outputs
PDF export for directly shareable requirements documents

Learnings

“Better requirements do not come from more text, but from better facilitation. AI delivers the most value when prompting, UX, and output structure are designed together.”

Tech Stack

Next.js 16React 19TypeScriptClaude APIPrompt EngineeringPDF ExportZustandshadcn/uiTailwind CSS v4

GitHub

05Case Study

Anni Platform

AI avatar companion for elderly care and support settings

Tech Stack

TypeScript · Node.js · Express

A TypeScript monorepo for a voice-oriented AI companion in elderly care. The platform cleanly separates backend API, web frontend, and shared domain models, showing how I translate socially meaningful product ideas into maintainable technical architecture.

Challenge

Conversational AI for sensitive user groups requires more than a chatbot. It needs clear system boundaries, a secure API layer, well-structured shared data models, and an architecture that can grow into speech, authentication, and operational requirements.

Solution

I set up a modular platform with an Express API, React/Vite frontend, shared package, Zod validation, Drizzle-based data access, and security middleware. That gives the product a clean technical foundation for later expansion.

Monorepo with apps/server, apps/web, and packages/shared
Express 5 API with Helmet, CORS, rate limiting, and JWT authentication
Drizzle ORM and PostgreSQL for type-safe data access
React 19 frontend with Vite for fast iteration
Shared types and utilities through a dedicated package
Claude SDK integration as a foundation for AI-assisted interactions

Learnings

“Early architecture work matters most when a product idea can evolve into a sensitive multi-part system. Good separation now reduces friction in every later expansion step.”

Tech Stack

TypeScriptNode.jsExpressReactViteDrizzle ORMPostgreSQLAnthropic SDKZodMonorepo

GitHub

06Case Study

QA Test Concept Skill

Knowledge product for ISTQB-aligned test concepts in the German market

Tech Stack

Prompt Engineering · Claude Skills · ISTQB

A reusable AI skill product that generates structured QA test concepts from project context, reviews existing documents, and checks them against ISTQB-required sections. The project shows how I combine QA methodology, prompt design, and reusable knowledge assets into a concrete deliverable.

Challenge

Many test concepts are either incomplete or too generic. In regulated or complex delivery environments, teams need structured quality documentation that is methodologically sound and still efficient to produce.

Solution

I built a skill with intake logic, document review, compliance checklist, and target-audience-specific outputs for public sector, banking, SaaS, and ERP contexts. Instead of generic AI prose, the result is a repeatable QA workflow with clear structure.

Coverage of all 10 ISTQB-required sections
Structured intake flow for new projects
Review logic with traffic-light scoring and compliance checklist
Output in Markdown or DOCX depending on context
Optimized for German IT projects in public sector, banking, SaaS, and ERP
Focus on reusable expert knowledge instead of one-off prompting

Learnings

“Not every valuable technical asset is a classic app. Skills and knowledge products scale expertise, reduce quality variance, and make domain knowledge operational.”

Tech Stack

Prompt EngineeringClaude SkillsISTQBQA GovernanceMarkdownDOCX Workflows

GitHub

07Case Study

Feierabendtrader

Swing trading breakout scanner with Yahoo Finance, Claude Haiku, and TradingView

Tech Stack

Next.js 15 · React 19 · TypeScript

A local Next.js app that delivers 5–10 justified swing trading breakout setups from the US stock market at the push of a button. The project combines free market data screening, AI-powered analysis with Claude Haiku (~$0.005 per scan), and TradingView charts in a lean, locally runnable tool — without auth, database, or hosting dependency.

Challenge

Manual stock screening is time-consuming and subjective. Existing tools are often expensive, subscription-based, or provide no explainable logic behind their setups.

Solution

I built a multi-stage screening funnel: Yahoo Finance filters ~80 candidates down to ~20, Claude Haiku analyzes these and returns the best 5–10 breakout setups with entry zone, stop-loss, risk/reward ratio, and justification. Everything runs locally, with no infrastructure overhead.

Next.js 15 app with React 19, TypeScript, and Tailwind CSS 4
Multi-stage screening funnel via Yahoo Finance (free, ~80→20 candidates)
Claude Haiku analysis for 5–10 prioritized breakout setups with reasoning
Setup cards with breakout level, entry zone, stop-loss, risk/reward, and TradingView chart
Market hours badge, localStorage cache (24h), and funnel metrics
Zod validation for type-safe API responses

Learnings

“AI-powered financial tools don’t have to be expensive. Targeted prompting on a solid data foundation is enough to turn ~80 raw candidates into qualified, explained setups in seconds — without any cloud dependency.”

Tech Stack

Next.js 15React 19TypeScriptTailwind CSS 4Claude HaikuAnthropic APIyahoo-finance2TradingView WidgetZod

GitHub

Articles about how I work

Short essays on how I connect QA, AI quality, requirements and delivery in practice.

View all articles

Mar 13, 20266 min read

What I have actually learned after several years in QA

Quality does not emerge during testing alone. It emerges in the system, in the process and in the way a team thinks.

QA StrategyTest AutomationQuality Systems

Read article

Mar 1, 20267 min read

How I built a SaaS without a traditional developer background - using AI and a QA mindset

NutriKompass started with a concrete operational problem in a therapeutic care setting and the question of how far AI, product thinking and QA experience can take you today.

SaaSAI WorkflowQA Mindset

Read article

Reusable Assets & Accelerators

Not just one-off projects: I build reusable assets that help teams reach robust outcomes faster.

Reusable AssetLLM QARelease GatesPlaywright

AI QA Release Gates

A reusable QA framework for LLM features with traceability, multi-model tests, and explicit release criteria before shipping.

Best suited for

Teams that want to qualify AI features beyond prototyping and make them measurable, regression-safe, and releasable.

Typical deliverables

Test suite for security, bias, RAG, performance, and UI
Requirements-to-test traceability
HTML reports and release gate logic

View repository

Reusable AssetRequirementsInterview FlowPDF Output

Guided Requirements Workflow

A structured interview flow that turns vague ideas into prioritized user stories, acceptance criteria, NFRs, and initial delivery structure.

Best suited for

Product owners, business teams, and delivery setups that need to move faster from unclear requests to implementable artifacts.

Typical deliverables

Guided discovery flow across 3 phases
Structured user stories and NFRs
Shareable PDF artifact for alignment and kickoff

View repository

Reusable AssetISTQBSkill ProductDOCX Workflows

QA Test Concept Skill

A knowledge and workflow asset for ISTQB-aligned test concepts, document reviews, and context-specific QA deliverables.

Best suited for

Organizations with heavy documentation needs, regulated contexts, or a need for methodologically sound QA structure.

Typical deliverables

Intake logic for new projects
Review and completeness checks for existing test concepts
Markdown or DOCX output depending on the target context

View repository

Capabilities & Technologies

The mix of QA, AI, requirements, and delivery that I use to make projects operationally reliable.

AI Quality Systems

Quality for both classic web products and LLM systems: from test strategy and evaluation logic to release gates.

LLM QARelease GatesPlaywrightRAG EvaluationSecurity TestingTraceabilityTest StrategyQA Governance

Product Engineering

Modern product stacks for production-oriented web apps, internal tools, and specialized SaaS products.

Next.js 16React 19TypeScripttRPC / APIsPrisma / DrizzlePostgreSQL / pgvectorTailwind / shadcnMonorepo Architecture

Automation & Delivery

Repeatable delivery with CI/CD, multi-provider AI, document-based outputs, and operational hardening.

GitHub ActionsPrompt EngineeringClaude / GPT / GeminiPDF / DOCX OutputsStripe / ResendNextAuth / Auth FlowsVercel / NetlifyOperational Hardening

Requirements & Leadership

Structure for vague requirements, complex stakeholder settings, and defensible project decisions.

Requirements EngineeringGuided DiscoveryStakeholder FacilitationScrum / KanbanPRINCE2 / IREB / ISTQBProduct ThinkingAgile DeliveryConsulting Leadership

Certifications

Verified credentials from internationally recognized bodies

2022

ISTQB Foundation Level

ISTQB

2023

Professional Scrum Master I

Scrum.org

2023

Professional Scrum Product Owner I

Scrum.org

2023

IREB CPRE Foundation Level

IREB

2024

PRINCE2 Foundation

AXELOS

How I Work

Principles that guide my work.

Quality as Mindset

Quality isn't a process step - it's a way of thinking. I integrate quality assurance into every phase, from requirements to release.

Product Thinking

I think in user problems and solutions, not features. Every technical decision must serve the product.

Ownership

I take responsibility for outcomes, not just tasks. When something doesn't work, I find a way.

Contact

Interested in working together? I'd love to hear from you.

or reach me directly

Email LinkedIn GitHub