Ridzwan G.

I add AI to SaaS products and web apps that are already running, and I take over builds that stalled: RAG and document Q&A, AI agents, and workflow automation with n8n. Laravel and Next.js are where I work every day. The clearest example: a client's Next.js app where pages took around five minutes to load and sometimes failed outright. The cause was a race condition in the data fetching on almost every page. Nothing in the codebase described how the app was supposed to flow end to end, so the tests had nothing to check against and the bug kept surviving. I traced the whole flow first, then fixed it. Those pages now load in seconds. Top Rated with 100% Job Success and 5.0 on every public review, and one client came back to hire me a second time. I have also worked on a multi-tenant SaaS where every business has its own users and one person can hold a different role in each, with financial reporting and forecasting on top. That one is under NDA, so I can walk you through the architecture but not show you the app. What I build - RAG and document Q&A: pgvector, OpenAI embeddings, page references the reader can check against the source PDF - AI agents and assistants: function calling with real tool access, streaming replies, booking and lead capture flows - LLM integration: Claude and OpenAI wired into an existing Laravel or Next.js app, keys and calls kept server side - Workflow automation: n8n connecting AI to WhatsApp, Google Calendar, Slack, Airtable and Google Sheets, plus Laravel queues, scheduler and Horizon when the work belongs inside the app - Fixed-price audit of an existing SaaS or web app that is slow, fragile, or that nobody is sure how it works. You get a flow map of how it actually behaves end to end, a ranked list of risks, and a fix plan with effort estimates. My own products, built end to end - Doxie AI: a RAG document Q&A SaaS on pgvector and OpenAI embeddings, answering with page references - SpreeBuddy: an AI agent for shopping, function calling with real tool access and streaming replies - Sanisched: an AI chatbot that takes a clinic booking, with n8n automation sending the WhatsApp reminder - Ridzone Leads: AI lead scoring on form intake, with n8n automation routing each lead to Airtable, Slack and email Most of the code is open. Stack - Backend: Laravel, Livewire, PHP, Node.js - Frontend: Next.js, React, TypeScript, Tailwind CSS - AI: Claude API, OpenAI API, embeddings, pgvector - Automation: n8n, Laravel queues, scheduler and Horizon - Database: PostgreSQL, Supabase, MySQL - Cloud and deploy: AWS (Lambda, S3, SQS, DynamoDB, API Gateway), Vercel, Docker How I work: scope and milestones agreed before I start, progress updates you do not have to ask for, end-to-end tests, and documentation left behind. Individual freelancer, no agency and no outsourcing. Available 30 to 40 hours per week, core hours 9am to 5pm GMT+7, happy to shift for calls. Message me with what you are building, where your data lives, and what a win looks like.