Manager - Data Science
The Insights and Decision Science (IDS) team is dedicated to enabling improved decision-making at Novartis by leveraging data and advanced analytics capabilities to generate actionable insights that drive business growth. We collaborate closely with the US business, bringing insights and challenging ideas to empower smarter, data-driven decision-making. The Gen AI team pioneers enterprise-scale Generative AI solutions, specializing in building production-ready LLM systems, multi-modal AI architectures, and sophisticated agentic workflows. We leverage cloud-native GenAI platforms (AWS Bedrock, Azure OpenAI) to deliver innovative solutions that transform unstructured data into actionable insights through advanced RAG systems, multi-agent orchestration, and vision-language models. Our team excels at rapid prototyping and delivery of cutting-edge GenAI applications while implementing robust evaluation frameworks, guardrails, and responsible AI practices to ensure scalable, ethical deployments across the organization.
This role is crucial for pioneering the safe and effective deployment of autonomous AI agents that can transform business operations while maintaining robust guardrails and evaluation frameworks. The Manager combines deep technical expertise in agentic AI with strategic leadership to ensure these powerful systems deliver measurable business value without compromising safety, ethics, or reliability. As organizations move beyond basic GenAI applications toward autonomous agents that can reason, plan, and execute complex tasks, this roles unique blend of hands-on technical skills and evaluation expertise becomes essential for navigating the opportunities and risks of advanced AI systems.
This role is crucial for pioneering the safe and effective deployment of autonomous AI agents that can transform business operations while maintaining robust guardrails and evaluation frameworks. The Manager combines deep technical expertise in agentic AI with strategic leadership to ensure these powerful systems deliver measurable business value without compromising safety, ethics, or reliability. As organizations move beyond basic GenAI applications toward autonomous agents that can reason, plan, and execute complex tasks, this roles unique blend of hands-on technical skills and evaluation expertise becomes essential for navigating the opportunities and risks of advanced AI systems.
About the Role
Location Hyderabad #LI Hybrid
Key Accountability
- Lead the design and implementation of sophisticated agentic AI systems, including multi-agent orchestration and autonomous workflows
- Define architectural standards for agent development, including tool integration, memory management, and inter-agent communication protocols
- Establish comprehensive evaluation frameworks for agentic systems, measuring task completion, reasoning quality, and safety compliance
- Design and implement guardrails for autonomous agents, including behavioral boundaries, output validation, and fallback mechanisms
- Drive hands-on prototyping of complex agent systems while providing technical oversight to development teams
- Create testing methodologies for agent reliability, including edge case handling, adversarial testing, and performance benchmarking
- Balance technical depth with strategic thinking to identify where agentic AI can transform business processes
- Lead technical reviews of agent architectures, personally debugging complex multi-agent interaction issues
- Implement LLMOps and agent lifecycle workflows, including versioning, prompt management, evaluation pipelines, and performance monitoring.
- Drive experimentation with emerging agent tooling, frameworks, and orchestration patterns to maintain a cutting-edge internal capability.
- Document reference architectures, reusable components, and best practices to accelerate agent development across teams.
- Conduct technical deep dives and capability assessments to ensure readiness of agentic systems for production deployment.
Essential Requirements
- 6+ years in AI/ML with maximum 1-2 years in generative AI and demonstrated expertise in agentic systems
- Hands-on coding experience with complex agentic AI implementations, including multi-agent systems and tool-use agents
- Deep technical knowledge of agent evaluation metrics task success rates, reasoning traces, tool-use efficiency, and safety violations
- Proven experience implem enting production guardrails content filtering, action limitations, human-in-the-loop systems
- Strong understanding of agent architectures ReAct, Plan-and-Execute, Reflexion, and emerging patterns
