Shirsha Chaudhuri

Shirsha Ray Chaudhuri is an enterprise AI and research-engineering leader who helped build Open Arena, Thomson Reuters’ secure generative AI playground. Her work confronts a central problem in enterprise automation: turning experimental models into reliable systems that operate across established software, professional workflows, and human decision-making.

Chaudhuri studied computer science at the Birla Institute of Technology and Science, Pilani, began her career at Oracle India, and held engineering roles across Motorola, Alcatel-Lucent, Tektronix Communications, and Nokia Networks. She subsequently studied business analytics and intelligence at the Indian Institute of Management Bangalore and worked at Mercedes-Benz Research and Development India on electric-bus route planning, predictive maintenance for truck-service operations, and AI-assisted field operations.

As a director of research engineering at Thomson Reuters Labs, she applied that industrial and telecommunications experience to professional-information systems and legal technology. She contributed to research on deploying and sustaining legal AI systems, with particular attention to the operational demands of production deployment.

In 2023, she coauthored an account of building Open Arena in under six weeks. The internal platform combined multiple language models, protected enterprise data, document question-answering, summarization, and retrieval-augmented generation, giving employees without programming backgrounds a practical environment for experimenting with generative AI.

What dependable enterprise automation requires

• Redesign complete workflows. Chaudhuri sees limited value in replacing disconnected tasks one at a time. Customer-support escalations can span service desks, IT operations, engineering, testing, and observability; editorial work crosses research, review, approvals, and publication. Her approach to enterprise workflow automation treats those handoffs and responsibilities as design requirements.

• Connect agents to operational reality. Useful automation depends on integrating existing enterprise software and legacy mainframes while assembling context scattered across logs, tickets, internal conversations, and business applications. Domain specialists must help determine which information and decisions actually matter.

• Make production guardrails nonnegotiable. Reliability, economic value, security, governance, deployment standards, and agent maintenance determine whether organizations can trust automated processes. Chaudhuri has highlighted production failures involving weak guardrails and misplaced confidence in Model Context Protocol integrations.

• Preserve human authority. She advocates human-in-the-loop approvals and a collaborative human-agent experience that keeps consequential decisions under professional oversight. Her leadership also emphasizes helping colleagues develop and advance.