Job
Description
Role OverviewThis is a strategic and hands-on role at the intersection of LLMs, agent-based systems, and enterprise AI engineering. You’ll design and implement autonomous agent workflows, enrich LLMs with structured knowledge, and architect tool-using agents capable of solving real-world business problems. Key ResponsibilitiesDesign, fine-tune, and evaluate Large Language Models (LLMs) for agent-based reasoning, natural language interaction, and workflow orchestration.Build autonomous agents using frameworks like LangChain, AutoGen, CrewAI, or similar, integrated with tool use, memory, and multi-step planning.Implement Agentic Frameworks and Digital TwinsDevelop and optimize Retrieval-Augmented Generation (RAG) pipelines using vector databases (e.g., FAISS, Weaviate) and hybrid architectures like GraphRAG.Engineer agent workflows with tool calling, multi-agent collaboration, and long-term memory (episodic + semantic).Monitor and improve performance metrics like reasoning accuracy, hallucination rate, and latency under production constraints.Collaborate with DevOps, product, and platform teams to scale agents using Docker, Kubernetes, and Azure ML. Must-Have Skills3–6 years of experience in LLM-based NLP / AI roles, with hands-on deployment experience.Proficiency in LangChain, AutoGen, CrewAI, or similar agent frameworks (real-world examples preferred).Strong understanding of agent memory, task planning, and function/tool calling workflows.Experience with RAG pipelines, embedding models, and vector search systems.Deep knowledge of Transformers, Hugging Face ecosystem, and custom prompt engineering.Proficiency in Python and ML libraries like PyTorch, TensorFlow.Familiarity with knowledge graphs, enterprise data systems, and orchestration strategies. Preferred QualificationsExperience designing persona-aware agents for enterprise use cases (BFSI, healthcare, public sector).Familiarity with BertGraph, GraphRAG, or similar graph-augmented architectures.Experience with Azure AI Studio, OpenAI APIs, Anthropic Claude, or Meta LLaMA models.Background in agent evaluation: using metrics for tool use accuracy, conversation consistency, and action success rates.
