Source description
About the role
Title: AI Engineer — Agentic AI & Supply Chain Automation Company: Nuvo AI (AI arm of Meril Life Sciences) Location: Vapi, Gujarat (on-site) Experience: 2–4 years About the Role We're building the next generation of AI-driven automation for Meril Life Sciences, one of India's largest medical device companies. You'll own the design and development of agentic AI systems that transform how our supply chain operates — demand forecasting, inventory optimization, supplier intelligence, procurement automation, and logistics. This is not a wrap an LLM around a chatbot role. You'll design multi-agent systems that reason over enterprise data, coordinate with ERPs and vendor systems, and take real actions with real business impact. What You'll Do Design and build end-to-end agentic AI systems for supply chain use cases (forecasting agents, procurement agents, supplier-risk agents, inventory-optimization agents) Architect multi-agent workflows using LangGraph, CrewAI, or AutoGen — pick the right tool for the problem, not the trendy one Build production-grade APIs and services in FastAPI/Python that expose AI capabilities to internal teams Own the full lifecycle: prompt engineering, retrieval design, evaluation, deployment, monitoring, iteration Integrate with enterprise systems (ERP, WMS, supplier portals) and design robust tool-use patterns Set up observability and evaluation pipelines using Langfuse, Phoenix (Arize), or LangSmith Work directly with supply chain domain experts to translate business problems into AI solutions Must-Have 2–4 years of hands-on AI/ML engineering experience, with at least 1 year building LLM-based systems Strong Python and FastAPI; comfortable designing REST APIs and async workflows Production experience with at least one agentic framework: LangGraph, LangChain, CrewAI, or AutoGen Solid grounding in RAG systems — chunking, embeddings, hybrid retrieval, re-ranking Experience with PostgreSQL and at least one of: Redis, Cassandra, or Neo4j Experience deploying LLM-based systems to production (any of vLLM, Triton, TensorRT, or cloud inference) Familiarity with LLM observability tooling (Langfuse / Phoenix / LangSmith) Ability to reason about latency, cost, and reliability trade-offs in LLM applications Good to Have Experience with supply chain, manufacturing, or enterprise B2B domains Knowledge graph experience (Neo4j) for supplier/product relationships Multi-agent orchestration in production (not just POCs) Fine-tuning experience (LoRA, QLoRA) on domain-specific data Familiarity with time-series forecasting (Prophet, neural forecasters) — supply chain forecasting is a core use case and Exposure to regulated environments (medical devices, pharma, healthcare)