Door categorieën bladeren
Ontdekken
Fiverr Pro
Nederlands
$
USD
I will test your ai chatbot, llm app testing, and perform a basic security audit
Oekraïne
Ik spreek Oekraïens, Russisch, Engels, Frans, Duits, Pools, Spaans
5 bestellingen voltooid
QA Engineer, Manual Automation Testing, Web, Mobile, API
I’m a QA Engineer specializing in manual and automation testing for web, mobile, and API products. I help teams release stable software by running functional, regression, smoke, exploratory, and UI te...
Over deze dienst
Deploying an AI chatbot, LLM application, or web tool?
Ensure your AI gives accurate, hallucination-free answers while keeping your user data and endpoints secure against vulnerabilities and prompt injections.
As a specialized QA Engineer, I perform comprehensive AI Chatbot Testing, LLM Evaluation (Prompt Red Teaming), and Baseline Security Audits for web apps, SaaS, and AI-powered platforms.
️ What I Test & Deliver:
- AI & LLM Functional Testing: Response accuracy, hallucination detection, intent recognition, context retention, edge cases, and conversational flow logic.
- Prompt Injection & AI Jailbreak (Red Teaming): Testing guardrails against prompt leaks, system prompt overrides, bias, and toxic outputs.
- Chatbot Integration Testing: UI/UX widget checks, API response latency, multi-turn dialogues across web and mobile platforms.
- Basic Security & Vulnerability Audit: OWASP Top 10 baseline checks (XSS, SQLi, sensitive data exposure, SSL/TLS checks).
- Actionable Bug Reports: Detailed findings with reproduction steps, evidence screenshots/logs, and security mitigation.
Focus Areas: OpenAI/ChatGPT bots, Claude, custom LLMs, RAG apps, Customer Support bots, OWASP, API security.
Testapplicatie:
Software
Ontwikkelingstechnologie:
C/C++
•
Go
•
Java
•
Node.js
•
Python
Apparaat:
PC
•
iPhone
•
Android telefoon
Mijn portfolio
Veelgestelde vragen
What is included in AI Chatbot & LLM testing?
I evaluate multi-turn conversation flows, verify tone and context retention, detect hallucinations, and run edge-case scenarios to ensure your bot handles unexpected user inputs gracefully.
How do you perform Prompt Injection and Jailbreak testing?
I act as a "red teamer" attempting to bypass your system prompts and safety guardrails (using adversarial inputs, indirect injection, and override attempts) to verify that confidential data or internal system instructions cannot be leaked.
What does the baseline security audit cover?
I run non-destructive baseline checks aligned with OWASP principles: checking for common front-end/API vulnerabilities (XSS, input sanitization flaws, sensitive data exposure, cookie/session security, and endpoint response validations).
What deliverables will I receive?
You will receive a structured PDF/Excel report with categorized issues (Severity: Critical/High/Medium/Low), proof-of-concept steps, screenshots/logs, and practical remediation recommendations.
Do you need my system prompt to test?
Not strictly, but it helps — knowing the bot's intended rules lets me test more precisely whether it can be tricked into breaking them. Without it, I test based on reasonable expectations for a chatbot in your domain.
