Browse categories
Explore
Fiverr Pro
English
$
USD
I will test your ai chatbot or llm app for wrong answers and unsafe replies
Playwright FinOps and QA Automation Engineer
I'm a QA engineer and Scrum Master with 9+ years of experience testing web, mobile and enterprise software. My main testing areas are finance, healthcare and agriculture software, where accurate data ...
Level 2
Has met high performance criteria and has a proven track record for meeting client expectations.
About this Gig
Is your AI chatbot, LLM app or RAG assistant ready for real users? I test it the way customers use it, so wrong answers, unsafe replies and broken flows are found before launch.
WHAT I TEST
- Wrong or invented answers, checked against your source content
- RAG accuracy: does it answer from your documents?
- Multi-turn context and conversation flow
- Prompt injection and jailbreak attempts
- Sensitive topics, tone and safe fallback replies
- Web chat screen on desktop and mobile
WHAT YOU GET
- Every scenario marked pass or fail, with severity
- Each issue with the prompt, the reply and screenshot evidence
- A short risk summary: ready to launch or not
- Premium: a re-runnable Playwright test pack to retest after every model or prompt change
HOW IT WORKS
1. Share your chatbot link, test access and source documents
2. We agree the scenarios and scope
3. I test and score each scenario
4. You get the report and a short walkthrough
I have 9+ years in QA, including work on a healthcare AI chatbot portal. Please use test data only, never real customer or patient data.
Testing application:
Web application
Development technology:
Node.js
•
TypeScript
Device:
PC
•
iPhone
•
Android mobile phone
My Portfolio
FAQ
Do you need real customer data?
No. Please use test data only. I never need real customer or patient data to test your chatbot.
Which AI apps can you test?
Chatbots, LLM apps and RAG assistants that I can use through a web link or chat screen. Share the link and test access before ordering.
Is this a security or compliance audit?
No. I test answer quality, safe replies and basic prompt injection. Formal security or legal compliance audits need a specialist.
What is a scenario?
One test conversation with an expected result, such as a question your users ask. Each is marked pass or fail with severity.
