Israr M.

🏆 TOP RATED PLUS | 100% JOB SUCCESS | 900+ HOURS ON UPWORK I don't just close tickets. I like problems where the solution isn't completely obvious yet. I'm a Senior Laravel/PHP developer building SaaS platforms, custom business systems, and AI-powered applications. A growing part of my work sits at the intersection of Laravel and AI: agents, MCP servers, RAG systems, and automation that connects platforms which were never designed to talk to each other. Recent work: a Laravel platform running 500+ scheduled jobs daily against 8 external APIs for 40+ client accounts; a reporting pipeline that replaced about a week of manual work with output in minutes; and two open-source Filament plugins listed in the official Filament directory (AI Visibility, AI Monitor, 2,500+ installs). ━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🔧 WHAT I BUILD - Laravel SaaS & multi-tenant applications - Laravel + Vue.js / React / Next.js front ends - Custom AI agents & agentic workflows - MCP servers and tools that expose your app safely to AI - RAG systems and chatbots trained on your own data - LLM integrations: OpenAI, Claude, Gemini - Filament admin panels & Livewire / TALL stack - REST APIs, webhooks & third-party integrations - Stripe / Paddle billing, subscriptions & webhooks - Legacy PHP & Laravel modernization - Queue, database & performance optimization ━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🤖 AI THAT DOES MORE THAN SEND PROMPTS I build agents that use custom MCP tools, query your application data, call APIs, trigger workflows, generate reports and coordinate actions across platforms. In practice: internal assistants wired into a CRM or SaaS, MCP tools behind real permissions, RAG chatbots over company documents, read-only support assistants answering from live data, multi-step workflows using different models for different jobs, and evaluation suites so agent output is tested against known-good answers, not judged by eye. How I build: I run a small crew of agents of my own. A planner writes the plan, an adversarial agent attacks it until it holds, I build with Claude Code and Codex, a second agent attacks the finished code, and a live-testing agent proves it works with Playwright. Static analysis, tests and style checks gate every change. The tools make me faster. The review is what makes that speed safe. ━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🔍 WORKING IN EXISTING CODEBASES Much of my work is inside production systems I didn't build. Before changing anything I map the architecture, the data flow, the business rules and the patterns already in use, so new work fits the system instead of fighting it. My longest engagement here ran 376 hours over eight months on exactly that kind of work. ━━━━━━━━━━━━━━━━━━━━━━━━━━━ If you're building something around Laravel, Vue/React, AI agents, MCP, RAG or automation, send me the problem and I'll tell you how I'd approach it before you commit to anything.