Full-Stack Developer for AI Agent

"We use an internal CRM and want to give our management team a clear, AI-powered view of the business. We need a developer who can quickly build a working prototype of an AI agent that processes CRM records and turns them into KPIs, available both on the web and on mobile. This is a small, one-time task, ideal for someone who builds prototypes fast with Claude Code or Codex. What we need 1) AI agent (core of the project) - Connect to our CRM through its API, read records (deals, contacts, tasks, activities) and process them on a schedule. - The agent should follow a decision loop: fetch data, analyze, decide which metrics or alerts matter, generate KPIs and short summaries, save the results. - AI model integration with an LLM (OpenAI, Claude or similar). Simple machine learning for forecasting or anomaly detection is a plus. - Expose the results through a small REST backend (API development), so the web and mobile apps can use the same data. 2) Web dashboard (Next.js) - A responsive web application with charts, KPI cards and filters (by period, manager, pipeline stage). - Front-end development with HTML5, CSS and Tailwind CSS. Clean, simple UI; no heavy design work needed. - Our current CRM admin panel is built with Vue.js. Some existing widgets may need to be reused or rebuilt, so Vue experience is a plus. 3) Mobile app (React Native, preferred) - An AI mobile app for management: the same KPIs, push alerts for important changes, and a simple AI chatbot to ask questions like “How did sales do this week compared to last week?” - We haven’t fully decided on the stack, but we’re leaning toward React Native. If you’d suggest something else, explain why. In your proposal, please describe: 1) How you would implement the agent’s decision loop. Give a concrete example. For instance: the agent detects that deals stuck in “Negotiation” have grown 30% week over week, decides it’s a risk, creates an alert and adds a short explanation to the dashboard. 2) Which tools you’d use for AI agent development (LangChain, LangGraph, OpenAI Assistants, a custom loop, etc.) and why. 3) Whether we can later add a RAG system to analyze our database, with examples of what it could do, such as: - “Why did we lose most deals in Q3?”, answered from deal notes and call logs - Finding similar past clients or deals for a new lead - Summarizing a client’s full history before a meeting - Answering internal questions about processes, based on our documents and CRM data Ideal candidate 1) Full-stack development experience (Node.js/Python backend, Next.js, React Native) 2) Hands-on experience with AI agents, automation and LLM API integration 3) Uses Claude Code or Codex daily and can deliver a working prototype fast 4) Can explain technical decisions in simple terms to non-technical managers "