Python backend engineer

Experienced Python Backend Engineer — Agentic AI & RAG Systems
Location: On-site
Type: Full-time
Team: AI Platform / Backend
About the Role
We’re building the next generation of Lurny’s AI platform — realtime voice tutors, RAG-powered learning, multi-agent orchestration, and autonomous tutoring systems. We’re looking for a strong Python backend engineer with deep hands-on experience in Agentic AI and RAG systems to maintain, scale, and expand our AI backend infrastructure. You’ll work on systems powering Live voice agents, RAG pipelines, media generation, and upcoming multi-agent workflows.
What You’ll Do
● Maintain and scale 2 production Python backends — Flask (REST) and Fast API + Web Sockets (realtime AI)
● Build agentic AI systems — tool-using agents, planner-executor loops, multi-agent orchestration
● Design robust RAG pipelines — ingestion, chunking, embeddings, retrieval, reranking, citations
● Integrate Gemini, Open AI, and Claude APIs for streaming, tool calling, and structured outputs
● Own Celery-based async pipelines for video, podcast, and PPT generation
● Improve AI observability, evals, latency, and hallucination reduction
● Work on Authentication &Authorization of apis
● Collaborate with frontend and Node.js teams on APIs and integrations
Must-Have Skills Core Backend
● 3+ years of Python experience
● Strong in Flask + Fast API (REST & Web Sockets)
● Celery + Redis, Mongo DB, Docker, API architecture
Agentic AI (Critical)
● Built production-grade agents with tool/function calling and multi-step reasoning
● Experience with frameworks like Lang Graph, Open AI Agents SDK, Gemini
● Strong understanding of agent loops, memory, state management, guardrails, evals, and orchestration
● Experience designing multi-agent systems
RAG & LLMs
● End-to-end RAG pipeline experience
● Hands-on with vector DBs (Pinecone , Chroma)
● Hybrid retrieval, reranking, grounding, citation, hallucination reduction
● Experience with Gemini, Open AI, and Claude APIs
AI Coding Tools (Required)
● Comfortable with Claude Code (preferred), Cursor or Antigravity
● Strong at prompting, reviewing AI-generated code, and spotting hallucinated APIs
Nice to Have
● MCP servers, Lang Smith/Langfuse/Phoenix
● Eval frameworks (Ragas, Deep Eval, Promptfoo)
● Experience on voice AI (Deepgram, Eleven Labs, Live Kit, Pipecat)
● GCP / Vertex AI experience
● Basic Node.js or React familiarity