Agentic AI Engineer - Assistant Manager
Company: 2005 STATE STREET BANK AND TRUST COMPANY - TAIPEI BRANCH
Location: Bangalore, India
Type: Full-time
Posted: 2026-07-27
About this role
Agentic AI Engineer (5–10 years of professional experience)
Role Description
We are seeking an Agentic AI Engineer with a strong combination of applied data science and AI engineering skills to design, develop, evaluate, and productionize enterprise-grade agentic AI applications. The role requires hands-on expertise in Python, machine learning, Generative AI/LLMs, prompt engineering, tool integration, and cloud-based deployment.
The Agentic AI Engineer will build intelligent applications capable of planning, reasoning, routing, retrieving information, invoking tools, and executing multi-step workflows. The role will work closely with data scientists, AI/ML engineers, software engineers, data engineers, and platform teams to translate business requirements into secure, reliable, scalable, and measurable agentic AI solutions.
Roles & Responsibilities
• Design and develop agentic AI applications using patterns such as planning, routing, reflection, tool/function calling, memory, and multi-agent collaboration.
• Implement AI/ML and Generative AI solutions using Python, with an emphasis on quality, reliability, efficiency, and scalability.
• Develop orchestration workflows that connect LLMs with enterprise APIs, databases, retrieval systems, and other tools.
• Design and optimize prompts, system instructions, tool descriptions, routing logic, and context-management strategies.
• Build RAG-based solutions, including document processing, chunking, embedding, retrieval, reranking, grounding, and context optimization.
• Apply data science techniques to agent development, including problem framing, hypothesis design, experimentation, error analysis, and iterative model improvement.
• Define and implement evaluation frameworks for agentic AI systems, covering task completion, answer correctness, faithfulness, tool-selection accuracy, trajectory quality, latency, cost, safety, and reliability.
• Create representative evaluation datasets, test scen...