Principal AI Engineer

Bristol Myers Squibb seeks a Principal AI Engineer in Hyderabad, India, to design and deploy production autonomous multi-agent systems across business domains. The role focuses on graph-based agent orchestration, LLM integration and scalable enterprise applications.

Responsibilities

• Architect workflows with 10+ nodes, conditional routing, task decomposition, sequential and parallel execution, collaboration between agents, and error recovery. Develop reusable agent-node libraries, extensible platform patterns, versioning and unit, integration and end-to-end tests.
• Build asynchronous FastAPI applications integrating PostgreSQL, Redis and enterprise services. Deliver real-time streaming through Server-Sent Events and WebSockets, plus RESTful and event-driven APIs.
• Integrate cloud LLM providers; manage versioned, templated prompts; persist conversation state through Redis checkpointing; and enable external tools and data through function calling and Model Context Protocol. Combine LLM reasoning with rules and statistical analysis to produce structured analytical outputs.
• Implement tracing, telemetry and evaluation of agent success, reasoning quality and output accuracy; optimize performance, token use and LLM costs. Address LDAP, SSO, access controls, data governance and Responsible AI standards.
• Work with data engineers, business analysts and UX teams; mentor junior engineers in async Python, agent design and LLMOps; and contribute to architecture reviews and documentation.

Qualifications

• Bachelor's or master's degree in Computer Science, AI/ML, Engineering, Data Science or a related discipline; 9+ years of software engineering experience, including 2+ years building and deploying production LLM-powered applications.
• Experience with agent orchestration frameworks, cloud LLM providers, production asynchronous web frameworks—particularly FastAPI—Docker, Git, CI/CD and AWS, Azure or Google Cloud. Expert async Python skills and experience with multi-step autonomous agents, state management, conditional logic, tool calling, MCP and agent-to-agent communication are required. Knowledge of ReAct, Plan-and-Execute, LLMOps, observability and prompt engineering at scale, along with analytical, problem-solving and communication skills for globally distributed teams, is expected.
• LangGraph is preferred among orchestration frameworks. Global life-sciences experience, particularly in the GPS functional area, and experience with offshore technical teams and international stakeholders are pluses. The posting describes several company work models but does not specify which applies to this role.