Door categorieën bladeren
Ontdekken
Fiverr Pro
Nederlands
$
USD
I will do ai application testing llm QA and chatbot testing
Software Testing Specialist, Manual, Automation and AI QA
Welcome to my profile! 👋
Hello! I'm a Senior Software Quality Assurance Engineer with 5+ years of experience helping businesses deliver reliable, user-friendly, and high-quality software.
I special...
Niveau 2
Voldoet aan hoge prestatiecriteria en heeft een bewezen staat van dienst in het voldoen aan de verwachtingen van de klant.
Over deze dienst
Is your AI app hallucinating, failing on edge cases, or vulnerable to prompt injections?
As a Senior Software QA Engineer (5+ yrs) specializing in AI and LLM Quality Assurance, I perform end-to-end testing on AI chatbots, SaaS wrappers, RAG pipelines, and AI agents.
What I Test:
- Hallucination & Accuracy (Factual alignment & groundedness)
- Prompt Injection & Jailbreaks (System prompt leaks, bypasses)
- RAG Pipeline Quality (Retrieval relevance, context noise)
- Conversational Flow & Memory (Logic loops, multi-turn state)
- UI/UX & API Integration (Console errors, timeout handling)
What You Get:
- Master Bug & Evaluation Report (Google Sheets/Excel)
- LLM Metric Scorecard (Faithfulness, Relevance, Safety)
- Annotated Screenshots & HD Loom Video Proofs
- Guardrail & Prompt Fix Recommendations
From GPT wrappers to complex LangChain/LlamaIndex agents, I help you launch a secure, reliable AI product.
Protect your brand order now or message me for a custom quote!
Leeftijdscategorie:
Volwassen
Educatie:
Hoger onderwijs
Platformtesten:
Website testen
•
Mobiel testen
•
Software testen
Apparaat:
PC
•
iPhone
•
iPad
•
Android telefoon
•
Android tablet
Taal:
Nederlands
•
Engels
•
Frans
•
Duits
Veelgestelde vragen
What is the difference between standard web QA and AI application QA?
Standard QA tests static logic and UI buttons. AI QA evaluates non-deterministic behavior, such as hallucinations, prompt injections, guardrail bypasses, context memory loss, and retrieval accuracy.
How do you test for AI hallucinations?
I run domain-specific ground-truth test cases, checking whether the model fabricates facts, cites non-existent sources, or contradicts provided context (especially critical in RAG apps).
What is prompt injection / jailbreak testing?
I act as an ethical red-teamer, attempting adversarial inputs to see if your AI can be tricked into breaking its persona, revealing system prompts, generating toxic outputs, or leaking PII.
Do you test custom RAG (Retrieval-Augmented Generation) applications?
Yes! I test whether your app retrieves the right documents, handles context noise, and accurately grounds its answers in your uploaded knowledge base.
Which AI testing tools and frameworks do you use?
I combine hands-on manual exploratory testing with industry-standard SQA tools like Promptfoo for red-teaming, DeepEval and RAGAS for metric scoring, and Postman for API validation.
Can you test AI apps built on Bubble, Webflow, or custom stacks?
Absolutely. I test AI wrappers and applications regardless of the tech stack—whether built on no-code tools (Bubble, Make) or custom Python/TypeScript frameworks.
What format will I receive the final report in?
You get an Excel/Google Sheets report with color-coded severity tags, reproduction steps, prompt/response logs, annotated screenshots, and a Loom video walkthrough.
Will my system prompts and proprietary data remain safe?
Yes. All system prompts, API keys, and test datasets are treated with strict professional confidentiality under SQA standards.
Can you recommend fixes for prompt or guardrail vulnerabilities?
Yes! Every bug or vulnerability logged includes actionable suggestions on how to modify system prompts, adjust temperature settings, or implement defensive guardrails.
Do you test mobile AI apps as well as web apps?
Yes, I test web applications, mobile AI apps (iOS & Android), web browser extensions, and standalone conversational chatbots.
