Full-Stack Developer for AI Creative Workspace (SaaS) – Frontend + Backend

Full-Stack AI Engineer – Multi-AI Orchestration Platform (OpenAI + Claude) with Web Research Engine Job description: About the project I’m building an AI creative workspace where people think, write, design, and build in one place. It has four modes (Spark Ideas, Design, Write, Build), saved conversation history, file/image support, and subscription plans for individuals, families, and businesses. It’s designed to be calm and accessible, especially for neurodivergent users. The next phase is the product's core: a multi-AI system that works like an AI team rather than a single chatbot. Different AI models and research tools work together behind the scenes to solve problems, research, verify information, and produce one stronger final answer. I’m looking for an experienced full-stack developer with strong AI/LLM experience to build this across the frontend and backend. How the system should work A user submits one input. The system routes it to the best AI models and tools; the models process it, share findings, review each other’s work, and the system returns one improved, verified final output. What you’ll build Multi-AI orchestration An AI router/orchestrator that automatically picks the best model or research tool for each task (task routing, workflow control, prompt management, input/output handling) Integration with the OpenAI API, Claude API, and other models (Gemini, Llama, Mistral, etc.) AI agents that can call and communicate with each other and share information A shared workflow where one model generates an answer, another critiques or improves it, and another verifies the result. A collaboration layer that compares responses, merges the strongest insights, and combines them into one final answe.r A verification stage that checks for accuracy, consistency, completeness, and valid citations before the answer is shown AI research engine Web crawling and scraping to collect publicly available information (news, academic papers, government data, public APIs), respecting robots.txt and site terms Real-time web research when current information is needed Source tracking (URLs, timestamps, credibility) so users can see where information came from AI analysis of research data to verify facts, add context, and improve answer quality Storage and memory Long-term memory store for user preferences, past conversations, project context, and reusable knowledge Research results database for collected data, sources, notes, and citations Input and output Processing for text, documents, images, code, and structured data Frontend and platform Workspace UI, chat interface with streaming responses, mode switching, and a clear view of sources/citations Accessibility (WCAG 2.1 AA), responsive/mobile layouts User accounts, authentication, and subscription plans (Plus, Family with up to 5 profiles, Business with team workspaces and admin controls), payments (e.g. Stripe) Performance, security, cost control for API usage, and deployment Tech stack [Fill in, e.g. Next.js / React, Node.js or Python (FastAPI), PostgreSQL, vector database (pgvector/Pinecone), Redis, Stripe, AWS/Vercel. Or: “Open to your recommendation.”] Requirements Proven full-stack experience with live SaaS products (please share links) Hands-on experience with OpenAI and Claude APIs, including streaming and tool/function calling Experience building multi-agent or LLM orchestration systems (e.g. LangGraph, LangChain, CrewAI, AutoGen, or custom) Experience with RAG, embeddings, and vector databases Experience with web crawling/scraping (Playwright, Puppeteer, Scrapy, or similar) and responsible data collection Strong backend skills: APIs, queues/background jobs, databases, authentication Awareness of LLM costs, latency, and rate limits Clear communication and regular progress updates Nice to have Experience with AI evaluation, fact-checking, or hallucination reduction Experience with document/image processing (OCR, PDF parsing, vision models) Experience building accessible products Project details Type: [Ongoing, starting with a paid first milestone] Budget: [$ hourly range or fixed amount] Timeline: [e.g. start ASAP; first milestone: router + two-model workflow in 3–4 weeks] Hours: [e.g. 20–40 hrs/week] To apply, please include: Links to 2–3 projects you built, ideally AI, multi-agent, or SaaS apps Your recommended tech stack for this system and why How you’d design the router and the generate → critique → verify workflow How you’d keep API costs and response times under control with multiple models Your availability and rate Start your proposal with the word “Shapes,” so I know you read this po.st