We are looking for a Backend Engineer, middle level or above. Our backend is an event-driven system of microservices that connects customer channels with AI agents in real time: from a WhatsApp message to streaming an LLM reply into the chat widget. The business layer is built with Node.js, the AI layer with Python and LangGraph. The most interesting tasks live where these two meet, so we need someone with hands-on experience in both languages.
About Intelswift
Intelswift is an AI-native platform for customer service automation. We build a product where AI doesn’t just answer questions, it does real work: * AI agents for chat and email * Voice AI agents * AI Copilot for support agents * Omnichannel support * Knowledge base and AI analytics
We move fast, avoid bureaucracy, and use AI in our daily work.
Tech side: microservices on Node.js (Express) and AI services on Python (asyncio). The AI layer is built on LangGraph, with LangSmith for tracing and LiteLLM for model routing; RAG runs on vector databases (Pinecone, Weaviate, Qdrant). Services talk through Kafka, Redis Streams, RabbitMQ and gRPC, realtime goes over WebSocket and SSE, data lives in MongoDB. Infrastructure: Docker, Kubernetes, GitHub Actions, AWS and Azure.
What you’ll do: * Build and develop Node.js microservices. * Design and evolve AI agents on LangGraph: agent graphs, multi-agent swarms, tool calling, RAG. * Trace and evaluate agent behavior with LangSmith: answer quality, latency, cost. * Tasks at the border of the two stacks: Kafka contracts and gRPC between Node.js services and Python workers. * Event-driven logic: retries, DLQ, idempotency. * Realtime for the dashboard and widgets: WebSocket, SSE. * Integrations: WhatsApp, Telegram, Instagram, Viber, Messenger, Stripe, MCP. * Deploy services to Kubernetes. Later on, help prepare the platform for on-prem installations and the move to local STT/TTS/LLM models.
What we expect: * 3+ years of commercial backend development, solid production experience with both Node.js and Python (one of them can be your main language). * Python: async stack (asyncio/aiohttp), our AI services are fully async. * Hands-on experience with LangGraph or a similar agent framework: agents with tools, state and memory. * MongoDB: indexes, aggregations (Mongoose). * Message brokers (Kafka and/or RabbitMQ): consumer groups, retries, DLQ. * Redis: pub/sub or Streams. * Realtime on the backend: WebSocket and/or SSE. * Basic understanding of gRPC and protobuf. * Kubernetes as a confident user. * English B2+, Ukrainian C2.
Nice to have: * Helm and experience with on-prem or self-hosted installations. * Experience with local STT/TTS/LLM models. * TypeScript, Temporal. * Voice tech: WebRTC, SIP, telephony. * Experience with billing systems (Stripe). * Experience with multi-tenant B2B SaaS.
Required languages English B2 — Upper Intermediate Ukrainian C2 — Proficient
What we offer: * Real impact on the architecture of a product used by real customers. * Work with modern AI technologies: multi-agent systems, LLMs, voice agents, RAG. * Remote-first culture with async communication. * Minimum bureaucracy, maximum ownership. * Flexible schedule with focus on results, not hours worked. * Paid vacation and sick leave. * Competitive compensation.
