Full-Stack Engineer – AI Agent Orchestration Platform (Python + Next.js)

About the Role We are building a platform for creating, deploying, and managing AI agents — not models. Users build agents that reason, use tools, follow workflows, and automate tasks. We need a full-stack engineer to develop agent orchestration logic in Python and deliver an agent-building interface in Next.js. Key Responsibilities Design and implement agent orchestration logic in Python (LangGraph, CrewAI, AutoGen, or custom state machines) Build APIs for agent execution, tool calling, memory management, and multi-agent collaboration Develop the Next.js frontend: visual workflow builders, tool configuration panels, agent test chat interfaces Integrate agent runtimes with external tools (APIs, databases, browsers, Slack, GitHub, etc.) Handle streaming agent responses (SSE, WebSockets) from Python backend to Next.js UI Implement session management, agent versioning, and execution logs Optimize for low-latency agent reasoning and tool execution Required Skills Python (Agent Focus): Python 3.9+; LangChain, LangGraph, CrewAI, AutoGen, or Semantic Kernel; ReAct, Plan-and-Execute, or multi-agent patterns; tool/function calling (OpenAI/Anthropic schema, custom executors); async Python; basic LLM reasoning knowledge (not training/fine-tuning) Next.js Full Stack: Next.js 14+ (App Router); TypeScript + React; real-time streaming UI; node-based UI libraries (React Flow, Blockly); API routes/server actions to trigger agent runs General: REST + WebSocket; Redis for job queues/state caching; Docker for agent sandboxing Nice to Have Agent evaluators (LLM-as-judge, trajectory scoring) Vector stores for agent memory (Pinecone, Qdrant, pgvector) Tool registries or marketplace features Open-source contributions to agent frameworks Deployment on Vercel + Railway/AWS/Azure Application Requirements Portfolio of relevant work (live demos, GitHub repos, or case studies) — required for consideration Applications without a portfolio will not be shortlisted