Farhan A.

I turn AI-generated and half-built code - from Cursor, v0, Lovable, Bolt - into production Next.js, Node.js apps & Laravel apps, with AI features and automations that stay reliable after launch. Where I come in: Cleanup: AI-generated or vibe-coded apps (Cursor, v0, Lovable, Bolt) audited, hardened, and turned into stable, secure, production-ready systems Dashboards, admin panels, internal tools - reporting, inventory, roles, exports, plus AI-assisted search and natural-language reporting where it adds real value APIs, auth, webhooks, payments, and MCP-style tool access - the glue between your app, other systems, and now AI agents too AI inside your product: OpenAI / Claude for chat, classification, summarization, RAG-based document Q&A, and tool-calling agents that take actions, not just answer questions Automation: event in, workflow runs, database / Slack / email / Stripe updated - via Node jobs, n8n, or AI agent workflows that decide the next step Observability: errors, retries, logs, plus AI-specific tracking - cost, latency, output quality - so nothing fails or burns budget silently AI API cost & speed: model routing (cheap model for simple tasks, strong model for hard ones), prompt caching, and streaming - smaller LLM bills, faster responses Server cost: AI-generated code that works but wastes compute - N+1 queries, missing indexes, no caching, oversized builds - hosting bills brought back to what the app should actually cost How I work: AI writes the first draft. I take responsibility for what goes live - scope, edge cases, security basics, and a clear definition of done. I work from a ticket list or milestones, not open-ended hours. Stack: React, Next.js, TypeScript, Node.js, Nest.js, Laravel , PHP PostgreSQL, MongoDB, AWS, Docker, Tailwind AI: OpenAI + Claude APIs Automation: webhooks, scheduled jobs, n8n Send me your repo, your broken preview, or the workflow that shouldn't need babysitting. I'll tell you what's realistic in the first week.