Rahul B.

Your app works in the demo. I find out what happens when real users, real data and real attackers touch it. I’m a QA lead with 10 years of testing web, mobile and API products for clients in the US, UK, Canada, Australia and Europe, including banking, insurance, healthcare and SaaS platforms. Today most of my work is on AI: applications built with AI tools, and applications that contain AI. What I test 🤖 AI, LLM & chatbot testing ✅ Hallucinations, wrong or inconsistent answers, context loss in multi-turn conversations ✅ Prompt injection, jailbreaks, data leakage and guardrail testing ✅ RAG accuracy against your source documents ✅ Voice agents: intent handling, latency, interruptions and fallbacks 🛡️ Pre-launch audits for AI-built apps (Lovable, Bolt, Cursor, Replit, v0) ✅ Missing authorization checks: can user A see user B’s data? ✅ Exposed API keys, open Supabase/Firebase rules, unsafe input handling ✅ Broken payment and webhook flows, duplicate charges, refund edge cases ✅ Clear go / no-go report before you share your app with real users 💳 Fintech, payments & SaaS QA ✅ Stripe, Razorpay and PayPal flows in test mode: declines, 3D Secure, renewals, refunds ✅ Subscriptions, credits, plan limits and role-based access ⚙️ Core QA ✅ Functional, regression, exploratory and UAT testing ✅ API testing (Postman, REST, auth and error cases) ✅ Automation with Playwright / Selenium [edit to match] ✅ Cross-browser and real-device mobile testing ✅ Performance basics: Core Web Vitals, Lighthouse, load checks What you get ========== A short test plan before testing starts, so you know exactly what’s covered Bug reports with steps, screenshots or video, and severity, logged straight into Jira, Linear, ClickUp or GitHub A summary after each round: what’s fixed, what’s risky, what to ship Recent results ============= ✅ Found an authorization flaw in an AI-built SaaS app that exposed one of customer records before launch. ✅ Cut regression time from [3 days to 6 hours] for a fintechclient through automation. ✅ Tested a healthcare & insurance chatbot for accuracy and safety before rollout to 4000 users. Why work with me =============== AI can write code and even tests, but the tool that wrote the code shares its blind spots, and nobody is accountable for its verdict. I give you tested evidence and a human sign-off. Not sure where to start? Send me your app link and I’ll share the top issues I see in the public parts of it, free.